Search

Bit-length of digital images is decided by the Analog-to-Digital (AD) converter. The bit-length expansion technique is necessary for high-quality displays such as the liquid crystal display (LCD) and the plasma display. Bit-length expansion is also necessary for low-bit length images, which are degraded by the pseudo-contour. In this paper, we propose a bit-length expansion method from M-level to N-level [N=3(M-1)+1]. This operation means that two quantization levels are introduced between existing levels. The bit-length expansion method results in the estimation problem of an error image between an M-level's image and an N-level's image. The error image is the pattern image which consists of three values. We show the estimation method of the three values pattern image and that the proposed method can remove the pseudo-contour of digital images through several application results. © 2009 Wiley Periodicals, Inc. Electron Comm Jpn, 92(10): 32–40, 2009; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecj.10105 [ABSTRACT FROM AUTHOR]

In the information and communications field, digital images have come to be widely used in mobile communications devices, and digital image processing is very important. Because of the different resolutions of the displays and output devices in each terminal, digital images undergo reduction and enlargement processing very frequently. Conventionally, reduction and enlargement processing is performed using an interpolation algorithm based on the sinc function. However, because the frequency band increases when digital images are enlarged, an enlarged image of sufficient quality cannot be attained through interpolation. In recent years, an enlargement method has been proposed which uses estimation of the frequency bands (high-frequency components) which increase with the enlargement of a digital image. The authors have also clarified interpolation algorithms using the preservation of step edge signals, coordinate point warping for generating peak signals, and signal amplitude biasing. They have implemented an interpolation algorithm which produces high-frequency components. In calculating the bias in this method, the smaller of the biases found to either side of the interpolation is selected. This choice is not necessarily optimal; the larger bias should be selected depending on certain conditions. For example, the selection of the larger bias makes it possible to preserve the precipitous changes in a step edge signal. This makes it possible to preserve step edge signals and generate peaks, without also using coordinate point warping, by selecting the optimal bias according to local information. The proposed interpolation algorithm using data-dependent biasing was compared to an interpolation algorithm using coordinate point warping and signal amplitude biasing. The proposed algorithm was found to have a similar performance and clearly shown to be a method with low computational complexity. © 2007 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 90(8): 18–28, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20306 [ABSTRACT FROM AUTHOR]

The YUV 4:2:0 format is a representative format for color images in international standards. In this format, both color difference components U and V reduce the number of pixels in the horizontal and vertical directions by one-half. Consequently, this reduction degrades the color difference components. At a relatively high bit rate, the degradation produced by reducing the number of pixels is often worse than the degradation that depends on the coding. Thus, processing to compensate for the degradation due to the format transform becomes necessary. The pixels of the color difference components reduced once on the encoding side are determined by interpolation on the decoding side. The typical interpolation methods are bilinear interpolation, bicubic interpolation, and interpolation by a finite impulse response (FIR) filter. When these interpolation methods were applied to interpolate the color difference components in the YUV 4:2:0 format, blur and lower saturation were produced in the colors in the color difference components returned to the original number of pixels. In this paper, we propose a method that uses information in the luminance component not having a reduced number of pixels to interpolate the color difference components in the YUV 4:2:0 format. Consequently, color blurring is prevented in the edges of the image, and an objective is to improve the saturation in the entire image. We based the proposed method on bilinear interpolation and bicubic interpolation, and formulated an interpolation method using the luminance component as an extended form. The proposed method and the bicubic interpolation are compared in both objective and subjective evaluations, and demonstrate the effectiveness of the proposed method. © 2006 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 89(9): 1–10, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20218 [ABSTRACT FROM AUTHOR]

When a multiscale digital image is output into a device with fewer simultaneous generation scales, such as two scales (black and white), halftoning is performed for approximate representation of the grayscale. A typical approach is the error diffusion method. Problems in this error diffusion method are the generation of regular patterns (worms) in the flat portion and of constant value regions not halftoned after variations where the signal varies suddenly. In this paper, the constant value region is investigated. A new method is proposed to avoid constant value regions. The constant value region is generated by diffusion of the quantization error with an inappropriate value for the forward scanning direction. In order to prevent this problem, the edge sections where generation of the constant value region is expected are eliminated by a smoothing filter. In this way, inappropriateness of the quantization error values for regions with different values in the forward scanning direction is suppressed. In order to prevent blurring of the input image, the filtering process is a data-dependent type that is limited to avoiding constant value regions. The proposed method is a simple one allowing avoidance of the constant value region without causing blurring by inserting a simple data-dependent prefilter. © 2006 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 89(5): 1–11, 2006; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20214 [ABSTRACT FROM AUTHOR]

Since the dimensions for digital images are set by the acquisition system, digital image enlargement is indispensable when the dimensions that can be displayed are larger in the display system than in the acquisition system. Conventionally, this enlargement process was performed by interpolation. The digital image enlargement corresponds to a larger Nyquist frequency, but the frequency band that is expanded during the enlargement cannot be considered by the interpolation scheme, and blur is produced in the enlarged image. The parts where the image blur stands out subjectively are primarily the signals forming the image contours having the step edge shape and the details of the image consisting of mountain- and valley-shaped signals when viewed microscopically. Therefore, the proposed method moves the coordinate points interpolated to preserve the discontinuities of step edge signals and decides whether the interpolated points are the vertices of the mountain- or valley-shaped signals, and then moves the amplitudes of the interpolated points for creating these peak signals up or down. In other words, we explain an interpolation scheme involving the warping and biasing that randomly moves the coordinate points and the signal amplitudes, and implement digital image interpolation with improved resolution. Furthermore, the parameter settings for the proposed method are both objectively and subjectively evaluated. © 2004 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 88(2): 10–21, 2005; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.20095 [ABSTRACT FROM AUTHOR]

Enlargement of digital images is equivalent to narrowing the sampling spacing. Hence, it is necessary for enlargement of the digital images to estimate the high-frequency components lost in the sampling by some means. Recently, an enlargement method accompanied by estimation of lost high-frequency components has been proposed. The present authors have also noted the Laplacian pyramid layer expression, which is a multiresolution representation of digital images, and have proposed an enlargement method to estimate lost high-frequency components. Based on the Laplacian pyramid layered representation, estimation of unknown high-frequency components is equivalent to estimation of unknown high-resolution Laplacian components. Since the layered expression of the Laplacian pyramid has band division encoding as the objective, the image size between the layers is 2 and the enlargement method based on this expression is also limited to 2. However, it is possible to obtain a layered expression with an arbitrary size ratio by varying the passband width of the Gaussian filter and eliminating the pixels corresponding to the bandwidth, in place of continuous use of the half-band Gaussian filter and downsampling of the usual Laplacian pyramid layered image. In this paper, based on this concept, an enlargement method with high-frequency component estimation that allows setting of an arbitrary enlargement rate is proposed. The effectiveness of the proposed method is clarified from abundant application examples. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(8): 41–51, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10018 [ABSTRACT FROM AUTHOR]

One excellent method for restoration of images with superposed impulsive noise uses the position information of the impulsive noise for the pixels used for processing and of the processing point. This method consists of two steps. In Step 1, the impulsive noise existing in the image is detected and the noise position image is formed. In Step 2, by using the noise position image formed in Step 1, the filter processing is carried out on the pixels degraded by impulsive noise only with the pixels not affected by the impulsive noise. An excellent procedure for Step 2, using an average filter (PSA filter), has been proposed. However, in the method with an average filter, a uniform weight is provided on all pixels in the filter window. Therefore, if the generation probability of the impulsive noise is increased, the window size cannot be increased even if the filter window size should be increased. Also, an individual image has directionality specific to the image. This directionality is not taken into account in the filter processing. In the present paper, in order to resolve this problem, a PSWA filter is proposed for the filter processing in Step 2 by using the weighted average filter with weights considering the distance from the process point and the directionality of the image. In order to appropriately apply the proposed filter, there are two parameters (window size N and parameter κ) to be set. From experimental investigation, it is found that these two parameters do not depend on the type of the original image but only on the generation probability of the impulsive noise. Further, from the generation probability of the impulsive noise, the method for providing these two parameters is presented specifically. The effectiveness of the proposed method is confirmed through several processing examples. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(3): 33–43, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1138 [ABSTRACT FROM AUTHOR]

Enlargement processing of digital images is usually carried out by means of an interpolation function obtained by simplification of the sinc function. However, in such an interpolation method, restoration of the high-frequency components lost in digitizing the images is not carried out at all. Hence, the enlargement causes blurring for images with many edges and detailed signals. For one-dimensional signals, Carrato and colleagues have proposed an interpolation method in which the discontinuities of the step edge signals are preserved while other signals are linearly interpolated. In this method, no blurring occurs in the expansion processing of strong frames of the images. However, Carrato's method has the following problems. (1) Only the step edges in the image signals are taken into account so that consideration of the detailed signals is insufficient. (2) Extension of the one-dimensional interpolation method given by Carrato to a two-dimensional interpolation method has a problem in interpolation at the center of the lattice. In this paper, a method is proposed to accomplish improvement of Carrato's method. With regard to problem (1), the interpolation not only is effective for the step edge signals but also preserves the apex of the peak and valley signals. With regard to problem (2), improvement is planned by directly deriving the interpolation value from the original signals. The proposed interpolation processing is accomplished by describing the processing sequence by a rule. The proposed method consists of two sets of fuzzy reasoning. The proposed method and the conventional interpolation method are compared from the numerical and visual perspectives and the effectiveness of the proposed method is clarified. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(3): 57–66, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1140 [ABSTRACT FROM AUTHOR]

The digitization and restoration of images that have been degraded by white or black blotches seen in motion picture films prior to the 1950s have attracted attention. Conventional restoration methods are composed of two processes. The first process is motion estimation by block matching (BM). The second process matches the current frame to the previous and next frames based on the motion estimates and filters in the temporal direction. Blotch detection accompanies the filtering in the second process. The BM used in the first process is average BM or sorted BM. Blotch detection in the second process employs Spike Detection Index-a (SDIa). In conventional BM, however, there are limits to the accuracy of motion estimation when blotches are present in the blocks used to estimate the motion in images corrupted by blotches. Furthermore, the accuracy of the filtering in the second process is affected by the accuracy of the motion estimation. And although blotch detection based on SDIa is a simple method, the accuracy of its blotch detection is a problem. To solve these problems in this paper, we create a blotch position image by using a median filter with an iteration constraint as the first stage, and obtain a virtual original image with the blotches removed by filtering based on the blotch position image. The accuracy of the motion estimation is improved by using the virtual original image, and the accuracy of the filtering is improved by using the blotch position image in the filtering. Based on these innovations, we propose a new method to obtain a restored image sequence having enhanced image quality. The effectiveness of the proposed method is clearly demonstrated through a variety of examples. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 85(10): 8–18, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1121 [ABSTRACT FROM AUTHOR]

The authors propose a new method for restoring images corrupted by high-probability impulsive noise at 10 to 50%. The strength (number of instances) of impulsive noise present in the area adjacent to a processing point is not uniform even in images that are corrupted by high-probability impulsive noise. In addition, when a processing point itself does not represent high-probability impulsive noise, outputting the input signal as is without performing median filter processing is vital to avoid image degradation. Given the above, one approach for restoring images degraded by high-probability impulsive noise is to first determine whether or not a processing point represents impulsive noise, and then perform median filter processing only when it is determined that impulsive noise is present. Second, when performing median filter processing, the amount of impulsive noise adjacent to a processing point must be estimated, the size of the filter window appropriate to that amount must be set, and efforts to preserve the signal must be made. In the proposed method the authors were able to unify three types of fuzzy inference. In addition, the proposed method is superior at handling images degraded by impulsive noise over a probability range of 10 to 50% when the fuzzy sets are fixed by setting the fuzzy sets in the three fuzzy inferences appropriately. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 85(9): 43–57, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1118 [ABSTRACT FROM AUTHOR]

We propose a method that uses multiple Wiener filters based on region decomposition to restore natural images degraded by Gaussian noise. Usually, a Wiener filter is designed to be optimal for the entire image. However, because natural images are highly nonstationary, an effective method switches to the optimum filter for each local region. In this paper, we consider a method that repeatedly decomposes into four equal parts, the four regions obtained by dividing an image into four equal parts, and design a Wiener filter for each region. As the regions become smaller, the desired Wiener filter is not necessarily obtained because of the drop in reliability of the correlation functions used to find the Wiener filter. Therefore, we first study to what extent the restoration accuracy of a known source image is improved by ideal decomposition that repeatedly decomposes only regions where the accuracy will improve. We then present a practical method that sufficiently improves the restoration accuracy even when the source image is unknown. We next merge similar decomposed regions and investigate how many classes are essential for the decomposition and these class boundaries. Thus, we verify that the proposed decomposition implicitly accounts for properties such as the contours in the image and clearly demonstrate the suitability of the proposed method. © 2002 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 85(8): 51–61, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1112 [ABSTRACT FROM AUTHOR]

In the enlargement of a digital image, it is required to estimate the high-frequency components, which were lost in sampling. As a method of meeting this requirement, the authors have proposed an image enlargement method using a high-resolution neural network (NN). The high-resolution neural network provided excellent results of image enlargement. There remained, however, the following two problems. (1) When the image is enlarged by the high-resolution NN, artifacts are produced in the smooth region of the enlarged image when the magnification ratio is increased as 4, 8, . . . . (2) There is a variation of approximately 10% in the accuracy of the enlarged result, depending on the kind of image used in the training of the high-resolution NN. In order to cope with such problems, this paper newly proposes the high-resolution multi-neural network (MNN), which is based on the local variance. The training procedure is shown. In the high-resolution MNN, two NN are used, corresponding to the region with a rapid change (high local variance) and the region requiring a smooth interpolation (low local variance), in the image for which resolution improvement is required. The weighted sum of the two NN outputs is defined as the result of enlargement. It is shown through application examples that the two problems in the high-resolution NN are remedied by using the high-resolution MNN. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 84(11): 61–69, 2001 [ABSTRACT FROM AUTHOR]

We propose a new model of texture images using morphological filters. This method is based on a linear predictive model [autoregressive (AR) model], which is a representative texture image model. This method embeds the frequency characteristics of the texture image in the AR model. In this paper, we will demonstrate a new model that represents the spatial features of a texture image. Morphological filters can remove the spatial features and are used for pattern spectra. In this paper, we consider embedding the spatial features in the structuring elements of morphological filters and propose an Opening Model (OM) to model the convex up features and a Closing Model (CM) to model the convex down features. We examine the reliability of the proposed model and present improvements to the low-reliability model. Furthermore, our recognition tests clearly show excellent recognition when OM and CM are used together. © 2001 Scripta Technica, Electron Comm Jpn Pt 3, 84(7): 53–62, 2001 [ABSTRACT FROM AUTHOR]

Since the enlargement of digital images is carried out by insertion of sample points, the Nyquist frequency of the enlarged image is naturally larger than that prior to enlargement. Hence, estimation of unknown higher-frequency components is required for enlargement of digital images. On the other hand, Laplacian pyramid representation is widely known as a layered expression of the images. Based on this Laplacian pyramid representation, the unknown higher-frequency components to be estimated at the time of enlargement of digital images correspond to the unknown higher-resolution Laplacian components. There is strong correlation between the Laplacian images at different layers. Specifically, the edge information spans several continuous layers as zero crossings and hence appears at the same location. Differences in the layers are recognized as differences in gradient near the zero crossing. In this paper, this property is directly used to provide a method of obtaining unknown higher-resolution Laplacian images from low-resolution Laplacian images (direct method). The results of enlargement by the direct method are particularly good in the smoothly varying parts of the images, but there are some difficulties in enlargement at the edges and of details. The paper proposes a hybrid method in which the direct method is combined with the method of Greenspan, with excellent enlargement results at edges and details, after the enlargement of images by estimation of the unknown higher-resolution Laplacian images as in the direct method. It is demonstrated by many application examples that this hybrid method is reasonably simple and is superior numerically and subjectively. © 2001 Scripta Technica, Electron Comm Jpn Pt 2, 84(6): 40–49, 2001 [ABSTRACT FROM AUTHOR]

In this paper, the authors propose a data-dependent weighted average filter (Video-DDWA: Video Data-Dependent Weighted Average) aimed at restoration of dynamic images deteriorated due to Gaussian additive noises. As proposed by the authors, this filter is based on the data-dependent processing, in which the filter weight is varied by multiple local information items derived from the data proximal to the processing point in a still image, with extension of this processing from a spatial filter to a temporal-spatial filter; and then the weight of adjacent frames is determined by detecting presence or absence of motion from the motion information such as new local information. There are several conventional methods of restoration of dynamic images that involve motion compensation with subsequent spatiotemporal filter processing, but they all have limitations as to the filter restoration performance due to noise-affected deterioration of the estimation accuracy of the motion vector. The proposed data-dependent filter using the motion information has among others the following advantages: (1) Because it involves detection of the motion degree at which noise influence is suppressed, it enables restoration of dynamic images featuring a high noise cancellation performance in still areas, without motion deterioration due to filter processing in motion areas—in other words, with processing that does not cause deterioration of movement and (2) the computing load is lower than that with motion-compensated filters. As compared to the conventional methods using motion compensation, this method enables attaining a high restoration performance not only for signals with a low S/N ratio at which the accuracy of the motion compensation estimation vector starts decreasing, but also for signals with a wide range of S/N ratios. The authors demonstrate with various application examples that this method is efficient for restoration of dynamic images. © 2000 Scripta Technica, Electron Comm Jpn Pt 3, 84(4): 1–10, 2001 [ABSTRACT FROM AUTHOR]

Since weighted median filters effective in recovering images belong to the class of stack filters, the authors have proposed a fuzzy weighted median (FWM) filter, which is a weighted median filter defined by a fuzzy Boolean function, and have presented its design method. The design of the FWM filter can be considered in terms of the design of the weight and the design of the fuzzy Boolean function separately. In the design methods developed to date, the fuzzy Boolean functions are sigmoid-type functions and the weights are S-type functions with only the distance from the processing point as a variable, which are approximated and require derivations of four parameters, which is an easy task. Such methods, however, are restricted to the class of designable FWM filters. In this paper, the class of designable FWM filters has been significantly expanded to incorporate nearly all classes of FWM filters by establishing all weights independently and by defining the fuzzy Boolean functions by appropriate functions in addition to the sigmoid-type functions. Thus, a high-performance filter is designed by the filter weight and the fuzzy Boolean function designed for the kinds of added noise for real image processing examples, which change more flexibly as compared with the FWM filters developed in the previous design methods. In addition, a design procedure for realistic cases in which only the images which are targets of processing are known is presented and a design method for an FWM filter, which is effective in practical terms, is presented by showing its performance. © 2000 Scripta Technica, Electron Comm Jpn Pt 3, 84(2): 95–106, 2001 [ABSTRACT FROM AUTHOR]

We propose a method to restore images degraded by additive noise by threshold decomposition filtering. The background regions in many natural images can be interpreted as concentrations of pixels having amplitudes close to certain constant values. These concentrations of pixels are observed as the peaks in a histogram. In this paper, we present a numerical algorithm to determine whether a two-channel filter bank for threshold decomposition is effectively applied to images having one histogram peak, in particular. We also explain how to determine specific thresholds for applicable images. Thus, the application of threshold decomposition is clearly indicated for images having at least one peak in its histogram. Then by applying the results of this study, we demonstrate through specific examples the precise application of threshold decomposition on images having different histogram characteristics. © 2000 Scripta Technica, Electron Comm Jpn Pt 3, 83(10): 17–26, 2000 [ABSTRACT FROM AUTHOR]

We have presented in this paper a new enhancement technique utilizing the high-frequency component information and the edge information sensor defined in Ramponi's method. These kinds of information were used in Ramponi's method, and it is clear that they were effective for the enhancement of noisy images, carrying out enhancement of edges only without enhancing the noise. However, enough extraction is not done because of the way the two kinds of information are considered. Therefore, we proposed in this paper a method based on fuzzy inference, in which the antecedent variables are these two kinds of information and the inference result is the enhancing component. The fuzzy sets for both kinds of information were established experimentally. Through application examples we made clear that we can set the fuzzy sets independently of the kind of images or the additive noise variance, and still greatly improve the enhancement results from Ramponi's method. In this paper we focused on the tuning of the enhancement function ƒ(i,j), whose variables are the edge information sensor defined by Ramponi (square of amplitudes) and the high-frequency component information. Naturally, the scheme of computation for the edge information sensor is still one subject of study, but we leave this topic for further investigation. © 2000 Scripta Technica, Electron Comm Jpn Pt 3, 83(7): 1–11, 2000 [ABSTRACT FROM AUTHOR]

Aiming at the elimination of Gaussian noise and impulsive noise, the authors have proposed the median filter and mean hybrid (MMH) filter, which combines the mean filter, which is useful in eliminating Gaussian noise, with the median filter, which is useful in eliminating impulsive noise. This paper proposes the learning MMH (LMMH) filter with the learning function, where the mean filter and the median filter in the MMH filter are replaced by the FIR filter and the order statistics (OS) filter, respectively, which are more general filters. The LMMH filter is optimized by learning, and the information concerning the position and the order in the training signal are well reflected on the LMMH filter, which improves the accuracy of the signal restoration, compared to the MMH filter. This paper further proposes a method by which the learning signal to be used in the learning of the LMMH filter can be constructed based only on the signal to be processed, which is degraded by the noise. By applying both the LMMH filter and the method of constructing the learning signal, a precise method for learning of the LMMH filter is provided, even in actual situations where the original signal for the signal to be processed is unknown. It is shown through an application example that the LMMH filter obtained by the learning exhibits a stable and better performance than the MMH filter. © 1999 Scripta Technica, Electron Comm Jpn Pt 3, 83(3): 95–107, 2000 [ABSTRACT FROM AUTHOR]

The stack filter is the largest class of nonlinear digital filters, including weighted median filters and morphology filters. The design of the stack filter is reduced to the problem of determination of the Boolean function defining its filter and is divided into design methods of estimation type using a training signal, and design methods maximizing the noise elimination capability under the constraint of preserving a specific configuration. With a view to expanding the class of stack filters, the authors have proposed the fuzzy stack filter, a stack filter defined by the fuzzy Boolean function. Also, for the fuzzy weighted median filter, as the most fundamental fuzzy stack filter, a design method based on the estimation standpoint is provided. In this paper, a new fuzzy stack filter preserving a specific structure is explained. The noise elimination characteristics of the fuzzy stack filter preserving a specific structure and the preservation characteristics of the specific structure under the effect of the noise are clarified. Further, a procedure is presented for deriving a filter suitable for restoration of the degraded images in a wide range without departing from the standpoint of a design method preserving the specific structure (for which empirical information on the signal to be processed cannot be obtained). The effectiveness of the filter obtained by this procedure is presented. © 1999 Scripta Technica, Electron Comm Jpn Pt 3, 83(2): 11–21, 2000 [ABSTRACT FROM AUTHOR]

When edges are to be extracted from an image signal with superposed impulsive additive noises, the noise must be eliminated beforehand. In other words, two independent processing operations must be executed, in the order of noise elimination and edge detection. The elimination of the impulsive noise naturally degrades the edges of the original signal. Consequently, noise elimination and edge detection should be executed in connection with each other. This study tries to realize a new edge detector method, which includes both impulsive noise elimination and edge detection by using fuzzy inference. In the proposed method, the extent of edge detection and the extent of noise elimination can be adjusted continuously by varying the setting of the membership function for the fuzzy inference. This paper presents a detailed guide for setting the membership function through several applied examples and demonstrates the usefulness of the approach. © 1999 Scripta Technica, Electron Comm Jpn Pt 3, 83(1): 61–69, 2000 [ABSTRACT FROM AUTHOR]

The stack filter is the most systematic among the nonlinear filters, including the weighted median filter and the morphology filter. The authors undertook to expand the class of stack filters and to increase the number of degrees of freedom in the design. They employed a fuzzy Boolean function in which the output Boolean function takes a continuous value between “0” and “1,” and proposed a fuzzy stack filter defined by that function. Up to the present, the fuzzy median filter and the fuzzy center weighted median filter have been presented as the most fundamental and simplest fuzzy stack filters. This paper proposes a fuzzy weighted median filter in which the weighted median filter, which is a more general nonlinear filter, is given a fuzzy property through the stack filter. The fuzzy weighted median filter changes continuously from the weighted median filter to the weighted average filter by varying the setting of the fuzzy Boolean function. This implies that the fuzzy weighted median filter has a high noise elimination ability for both impulsive and Gaussian noise. The fuzzy weighted median filter can be designed independently for the weight function and the fuzzy Boolean function. This paper aims at a simple design for the weight function and the fuzzy Boolean function, without losing its general character. The weight is approximated by an S-shaped function with only the distance from the processed point as the parameter, and the fuzzy Boolean function is approximated by a sigmoid function. The design is reduced to the determination of four parameters, which is performed by the LMS algorithm. Through application examples, the properties and the effectiveness of the fuzzy weighted median filter are revealed. © 1999 Scripta Technica, Electron Comm Jpn Pt 3, 82(10): 40–49, 1999 [ABSTRACT FROM AUTHOR]

Nonlinear filters are known to be effective for restoration of image signals subject to additive noise. Among them, the most typical are median filters, and particularly popular are stack filters. Signal processing in the stack filters involves the following three stages: threshold value decomposition, calculation of the Boolean function output at each level, and signal synthesis. The Boolean function is defined by the binary input/output. However, to achieve better filter characteristics and design diversity, the authors considered using fuzzy Boolean functions defined by continuous output values from 0 to 1, rather than conventional Boolean functions. In the current study, the authors performed fuzzification of such common stack filters as the median filter and center weighted median (CWM) filter. Fuzzification was performed by using fuzzy Boolean functions to define the median filter or the CWM filter. The fuzzy Boolean function is derived from fuzzy inference. As a next step, the authors developed a method for fuzzy inference optimization (tuning method). © 1998 Scripta Technica, Electron Comm Jpn Pt 3, 82(2): 83–94, 1999 [ABSTRACT FROM AUTHOR]

The enlargement of a digital image is a process of narrowing the sampling interval. Since the image before the enlargement and that after the enlargement have different Nyquist frequencies, it is necessary to predict or estimate the high-frequency components that are lost in the image before processing and to complement these components in the enlargement process. It is noted that the image can be represented as the sum of Gaussian components and Laplacian components in a hierarchical structure (pyramid representation). There exists a strong correlation between the Laplacian images of the layers. Based on that property, Greenspan and Anderson have proposed a nonlinear procedure where the unknown Laplacian image of high resolution (i.e., containing the lost high-frequency components) is estimated from the Laplacian image of low resolution. This paper notes that the Greenspan procedure can be network-structured and proposes an enlargement method for digital images using a neural network (NN) for improving the resolution, based on generalizing the network. It is shown that the enlarged NN can be constructed almost independently of the kind of training image or of the hierarchy (resolution). It is also shown that a better enlarged image can be derived compared with Greenspan's method. In other words, the effectiveness and practical usefulness of the proposed method are demonstrated. © 1998 Scripta Technica, Electron Comm Jpn Pt 3, 82(1): 19–29, 1999 [ABSTRACT FROM AUTHOR]

The authors have previously proposed a fuzzy stack filter that is defined by a fuzzy Boolean function. We have also presented a specific design method (called the fuzzy Boolean function determination method) for the most fundamental fuzzy stack filter, the FCWM (fuzzy center weighted median). By varying the fuzzy Boolean function, we can achieve both an arbitrary CWM filter and a weighted average value filter in which only the weights at arbitrary processing points are varied. Hence, in comparison with the CWM filter, our filter is better at eliminating Gaussian noise. In this paper, to improve the reconstruction accuracy of nonsteady state signals, such as the image signals, a data-dependent FCWM (DD-FCWM) is proposed anew. The data-dependent filter is an adaptive filter in which the filter parameters are varied in response to the nature of the data near the processing point (local information). In a DD-FCWM filter, the fuzzy Boolean function is varied by using local information. We demonstrate that the proposed DD-FCWM filter uses the local signal information to appropriately vary, not only the weights of the processing points, but also the filter mode (average value type or median type). Furthermore, by use of image processing examples, we demonstrate that highly accurate signal reconstruction can be attained. © 1998 Scripta Technica, Electron Comm Jpn Pt 3, 81(8): 27–40, 1998 [ABSTRACT FROM AUTHOR]

With a view to making efficient use of frequency and to prepare for full-scale mobile telephone service in the future by means of enclosing the electromagnetic waves in a small region, the basic propagation characteristics are studied for the case where a low base-station antenna (with the height of less than 15 m) is installed on the street of a metropolitan area. This paper proposes the two-wave model consisting of the diffracted wave traveling through a trench-shaped cross section surrounded by buildings and a road, and of the reflected wave from buildings. The propagation characteristics are studied theoretically and experimentally and the following results are obtained: 1) if there is no obstacle on the street, the diffracted wave is the principal contribution. The propagation loss is almost equivalent to the case of free-space transmission if the street width W and the radius R1 of the first maximum Fresnel zone are such that W⩾2R1 while it is larger than the latter for W⩽2R1 . The level change within the trench-shaped cross section is 5 dB (theoretical value); 2) if there are obstacles such as vehicles, the effect of the reflected wave is stronger than the diffracted wave if the receiver is running in close proximity to the vehicle while the diffracted wave becomes stronger if the distance is more than about 50 m. Under the street condition in which various vehicles are mixed, the additional loss (increase of the loss over the free-space loss) is about 15 and 10 dB, respectively, for the transmitting antenna height of 5 and 15 m at the distance of 1 km. Frequency dependence of the additional loss is not significant. [ABSTRACT FROM AUTHOR]

When a nonstationary signal containing sudden changes, such as an image signal, is degraded by an additive noise, a powerful means to recover the signal is to use a nonlinear filter. This paper uses a layered neural network, and proposes a new method to construct a nonlinear filter for restoring a signal corrupted by mixed noise (i.e., mixed noise composed of Gaussian noise and impulse noise). As the first step, a prototype filter is proposed, which is a combination of a median filter and a linear (averaging) filter. Then, it is shown that the prototype filter can be represented by a network structure. By interpreting the network representation by a layered neural network, the idea is extended to the median neural network hybrid (MNNH) filter. The MNNH filter can be trained by the back-propagation algorithm. The prototype filter already has a high mixed noise elimination performance, but the MNNH filter can further improve the performance by reflecting information on the signal to be processed (the original image and the additive noise) when such information is given. Lastly, the usefulness of the MNNH filter is demonstrated through various application examples. © 1998 Scripta Technica. Electron Comm Jpn Pt 3, 81(6): 52–60, 1998 [ABSTRACT FROM AUTHOR]

When a signal containing rapid changes, such as an image signal, is degraded by additive noise, a weighted median filter, which is a generalization of the median filter, is useful in its restoration. The weight of the weighted median filter is designed by using training signals, and is designed to be optimal for all training signals. The designed filter, however, cannot be optimal for local training signals. From that viewpoint, this paper presents a data-dependent weighted median filter in which the weight is made variable according to local information from the signal to be processed. A data-dependent weighted filter is of course desirable for highly accurate restoration of signals with mixed properties, as in image signals. In order to realize a high performance data-dependent filter, it is crucial to know accurately the local properties of the signal and to reflect them in the weights determined. In this study, three kinds of local information are used to represent the local properties of a signal, and a method for the derivation of weights using fuzzy inference is established. A data-dependent weighted median filter is constructed employing the proposed method. Its performance and effectiveness are examined in several application examples. © 1998 Scripta Technica. Electron Comm Jpn Pt 3, 81(4): 21–32, 1998 [ABSTRACT FROM AUTHOR]

An optimal filter in terms of removing noise can be determined for any signal distribution by using an order statistics (OS) filter, a typical nonlinear filter. For restoration of signals degraded by nonstationary signals such as an image signal, nonstationary noise, and mixed noise, an adaptive process in which the filter coefficients are varied in accordance with the local information is indispensable to improving signal accuracy. From these facts, an adaptive OS filter can be considered effective for image restoration. Most of the adaptive OS filters proposed so far have several fixed coefficients which are changed in accordance with the local information. However, for the restoration of signals such as deteriorated image signals, it is ideal to have the adaptive OS filter coefficients continuously change. In this paper, filter window signals are fuzzy-clustered into five predefined typical classes using a fuzzy inference system where three types of local information are the parameters of the antecedent part. Since there is an optimal OS filter for each of the five classes, we obtain the sum of outputs of the proposed filters from the filter output value which is optimal for each class, weighted by the degree of assignment of each filter class. With the proposed adaptive OS filter, the filter coefficients are varied continuously, and improvement of image restoration accuracy is attempted. © 1997 Scripta Technica, Inc. Electron Comm Jpn Pt 3, 80(9): 70–80, 1997 [ABSTRACT FROM AUTHOR]

The nonlinear filter is useful in restoring an image which is degraded by additive noise. The α-trimmed mean filter is a generalization of the median filter. In the α-trimmed mean filter, the data within the filter window an examined and a specified number of samples is selected as close to the median value from the ordered statistical viewpoint. Then the mean of the selected samples is output This mechanism is powerful in eliminating the impulsive noise and suppressing the non- impulsive noise. On the other band, it should be noted that the image and other signals are composed of various components (such as the flat part and the edge) and it is indispensable that the number of samples to be avenged in the α-trimmed mean filter should be made variable according to the local information (i.e., adaptive filtering) to realize a highly precise restoration of the signal. [ABSTRACT FROM AUTHOR]

This paper presents a design method of data-dependent filters that uses simplified fuzzy interference. Since the antecedents of fuzzy inference can comprise several local characteristics (i.e., observation values), it is possible for the fuzzy filter to adjust its weights to adapt to local image data. The tuning of membership functions and fuzzy rules of the proposed filter results in a least mean square (LMS)-like algorithm. Thus, local characteristics can be increased for the proposed fuzzy filter optimally. This paper, introduces a new observation value (calculated from local statistics) into the proposed filter. The proposed filter changes filter behavior according to the local properties of signals and provides good noise attenuation in all regions of image, including detail regions while still preserving the details. [ABSTRACT FROM AUTHOR]

When a random noise is superposed on a signal with an abrupt change (i.e., edge), as in the case of the image signal, there is a large overlap of the frequency component between the abrupt changes (edge) of the signal and the noise. Then it is impossible to realize both the noise elimination and the edge preservation sufficiently by the conventional linear filter, such as the mean filter. Thus, a nonlinear technique is required in order to eliminate the noise and preserve details of the original signal where the coefficient of the smoothing filter is determined according to the properties of the data in the window. There can be more than one observed value in the filter window that can be used to specify the properties of the data. It is difficult, however, to represent precisely the properties of the image signal using a single observed value. In addition, the observed values contain ambiguities since they are calculated from the local image data with superposed noise. In other words, it is required to determine the filter coefficients based on multiple observed values containing ambiguities. For such a problem, the fuzzy control rule, which is utilized effectively in control problems, can be applied to calculate the control variable (i.e., filter coefficient) adequately by the nonlinear technique from the multiple observed values containing ambiguities. In this paper, the nonlinear filtering technique based on the fuzzy control rule is presented and the usefulness of the filter in the noise elimination from the image signal is demonstrated. [ABSTRACT FROM AUTHOR]

A smoothing filter is applied to an observed signal in which the noise is added to the original signal. In the case of the mean filter, among smoothing filters, when the window length (neighborhood interval for smoothing) is large, the noise is suppressed sufficiently but the intermediate gray-level (blur in the case of an image) is produced near the edge. Conversely, when the window length is small, the noise suppression is insufficient. From such a viewpoint, an adequate definition of the window length is important. This paper considers especially the one-dimensional mean filter, and proposes a new method to define the window length in an adaptive way. The method is to examine the properties of the original signal from the observed signal in the neighborhood of the point to be processed, and to determine whether the window lengths is to be set equal on both sides of the processing point or to be set independently. Based on such a decision, the optimal window length is determined. The proposed method can cope with any kind of signal in a flexible way, leading to a satisfactory smoothing. By applying one-dimensional mean filters in this paper to four directions, an extension to the two-dimensional case is made easy for application to the image data. [ABSTRACT FROM AUTHOR]

As a generalization of a one-dimensional (ID) AR (Autoregressive) lattice filter structure, Parker et al. proposed a two-dimensional (2D) AR lattice filter with three reflection coefficients in each stage for the linear prediction of a given random signal. This paper shows that Parker's lattice model is not sufficient to realize any 2D AR process. Hence, we will use a more general lattice filter whose stage has six reflection coefficients. The determination procedures of reflection coefficients, and the algorithm for deriving an AR transfer function, a sufficient condition of filter stability, are considered. In the example shown, the proposed model is seen to be more general than that of Parker's. Finally, a new power spectral estimation method using the proposed lattice filter is considered. [ABSTRACT FROM AUTHOR]

ggThis paper discusses a design of two-dimensional IIR digital filter in the frequency domain, where the magnitude and the phase responses are assigned separately to the denominator and the numerator, respectively, and are designed independently, The design procedure is as follows. Considering only the phase, the all-pole filter (denominator polynomial) is designed. Then the numerator is designed from the 2-D mirror-image polynomial as a correction for the magnitude. In earlier methods of this kind, a nonlinear problem is encountered in the design of the all-pole filter in approximating the phase. By contrast, in this paper, the 2-D group-delay (phase) approximation is reduced to the 2-D spectral approximation by using the cepstrum, and the derivation of the parameters is reduced to the solution of the 2-D normal equation. From the viewpoint of ensuring the stability, an approximation by three-term separable all-pole filter is also shown, which makes the stability test easy. An algorithm is shown which solves the approximation problem for this case by iteratively solving the linear problem. The proposed design method features smaller computational complexity, since the denominator and the numerator polynomials are derived independently. In addition, the design is made easy since the linear problems arc handled throughout the filter design. [ABSTRACT FROM AUTHOR]

SUMMARY Frequency response analysis (FRA) is a powerful technique to detect internal abnormalities within power transformers. However, diagnosis criteria of power transformers by FRA have not been fully established yet. In this paper, a new evaluation method of degree of identification between two transfer functions is proposed. In the proposed method, a frequency window is employed and cross-correlation factor is calculated with moving the window. Furthermore, the proposed method is applied to transfer functions of real transformers with or without some abnormalities. Based on this discussion, objective criteria for diagnosis of abnormalities of transformer windings by FRA are proposed. The proposed criteria are based on a few data of transformers with some abnormalities. Therefore, the proposed criteria are tentative and they would be improved by accumulation of further data. [ABSTRACT FROM AUTHOR]

SUMMARY Soundness confirmation of internal windings of power transformers is important for reliable operation. In urgent situations such as invasions of lightning surge and external short-circuit faults, it is important to diagnose turn shorts and/or deformation of winding. In this paper, criteria in diagnosing turn shorts or heavy deformations of transformer windings using frequency response analysis (FRA) are proposed. In FRA, measured transfer functions are compared with past measured data in principle. However, past measured data are not necessary in proposed diagnosis criteria. Therefore, proposed criteria are useful in soundness confirmation of transformer winding in the urgent situation. [ABSTRACT FROM AUTHOR]

We present an electrochemical‐lossy mode resonance (LMR) sensing method that detects refractive indices and electroactive species. The LMR peaks of indium‐tin‐oxide in the transmittance‐wavelength spectra were significantly shifted as the applied potential between 1.0 and −0.5 V at 209 nm/V. The modulation was exploited for sensing the refractive index and electroactive species (ferrocyanide and methylene blue) in two ways: peak‐wavelength tracking and potential scanning. The potential‐scanning technique produced clear potential LMR peaks in the transmittance‐potential spectra for the first time, which were corresponded to the external refractive index. Meanwhile, the limits of detection of ferrocyanide and methylene blue were 7.5 and 25.3 μM, respectively, in peak‐wavelength tracking and 18.2 and 20.8 μM, respectively, in the potential scanning technique. [ABSTRACT FROM AUTHOR]

Histogram equalization (HE) is a simple and effective method for contrast enhancement as it can automatically define the gray-level transformation function based on the distribution of gray levels included in an image. HE fails to produce satisfactory results for a broad variety of low-contrast images because HE does not use the spatial features included in the input image. This paper proposes a novel contrast enhancement method using not only the conventional histogram but also the differential gray-level histogram of the input image. The differential gray-level histogram contains edge information from the input image. Edge information is one of the important spatial features of the image. We can extract the regions that are to be emphasized by using both the conventional histogram and the differential gray-level histogram. The input histogram is modified by the histogram of the extracted regions. The gray-level transformation function is derived using the modified histogram. [ABSTRACT FROM AUTHOR]

When a texture is considered as a set of granular components of various sizes and shapes, it is meaningful to determine its features in the spatial domain. It is known that such granular features in the spatial domain can be calculated by means of the pattern spectrum obtained by morphology operations. The pattern spectrum has the property that a peak exists at the point where the structuring element of the morphology filter agrees with the granular feature composing the texture. Utilizing this property, this paper proposes a method by which the features of the granules (shape and size) composing the texture are represented by a structuring element on the basis of pattern spectrum information. The proposed method restricts the textures to be analyzed to textures composed of granules of a single shape and size. A method of using the obtained model to decide whether the given texture can be analyzed is presented. With this approach, it is possible to decide whether or not the derived model represents the texture feature. © 2004 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 88(2): 31–40, 2005; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10070 [ABSTRACT FROM AUTHOR]

A filter using fuzzy inference has been proposed as a noise reduction filter for images with superposed mixed noise and its effectiveness has been demonstrated. When fuzzy inference is used, the optimization of its parameters is often a problem. For instance, Takashima and colleagues have proposed a configuration consisting of a cascaded connection of fuzzy controllers for impulsive noise and Gaussian noise. It is difficult to carry out optimization of both fuzzy controllers at the same time. In this paper, an optimization method is proposed for fuzzy controllers cascaded or connected in a layered manner. This method is applied to the fuzzy controller in the edge-preserving nonlinear filter proposed by Takashima. In this way the structure is extended to a tunable filter. From the simulation results, it is found that the proposed method is effective for performance enhancement of the filter and is also effective for adjustment of fuzzy controllers connected in a layered manner. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(5): 76–88, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.1150 [ABSTRACT FROM AUTHOR]

Background: Bevacizumab improves symptoms via reducing the peritumoral edema and/or normalizing blood brain barrier, and occasionally via reducing the tumor size. However, the effect against active cystic components has not been documented yet. Materials and Methods: Between 2008 and 2018, 139 patients with primary or metastatic brain tumors were treated with bevacizumab (BEV) in our institution. The images and symptoms before and after administration of BEV were examined, and changes in size of cysts were evaluated as follows: CR (complete disappearance), PR (reduction by ≥50%), MR (reduction by ≥25%), SD (size change <25%), PD (increase by ≥25%). The effect of BEV on tumor itself was determined according to Response Assessment in Neuro‐Oncology criteria. Results: Of the 139 patients, 21 (15.1%) had cystic components. The best responses of cysts to BEV treatment were as follows: CR 6, PR 7, MR 4, SD 4. The group of patients with progressively increasing cysts prior to BEV treatment had significant cyst size reduction compared to stable cyst size groups, at initial imaging after BEV (mean 62.6% vs 22.5%, P = .0055) and at best response timing (mean 76.3% vs 32.8%, P = .0050). Patients with cysts showed significant improvement in symptoms after the treatment with BEV compared to patients without cysts (P = .0033). However, response rate was not different between patients with or without cysts. Overall survival after starting BEV was not different between glioblastoma patients with or without cysts. Conclusion: Bevacizumab is effective against progressively increasing cysts. Although cysts reduction effect and tumor response and/or overall survival are independent, BEV may be effective in patients who are symptomatic due to cyst enlargement. [ABSTRACT FROM AUTHOR]

Cover Picture: Electrochemical Lossy Mode Resonance-based Fiber Optic Sensing for Electroactive Species (Electroanalysis 3/2023) Keywords: lossy mode resonance; potential scanning; refractive index; electrochemical redox EN lossy mode resonance potential scanning refractive index electrochemical redox 1 1 1 03/15/23 20230301 NES 230301 The cover image is based on the Research Article I Electrochemical Lossy Mode Resonance-based Fiber Optic Sensing for Electroactive Species i by Takuya Okazaki et al., DOI: 10.1002/elan.202200089. Lossy mode resonance, potential scanning, refractive index, electrochemical redox. [Extracted from the article]

Aim: The purpose of the present study was to elucidate the change of D-dimer and the possibility of deep vein thrombosis screening by D-dimer during pregnancy. Methods: One thousand, one hundred and thirty-one pregnant women were enrolled in the study from April 2006 to March 2007. D-dimer was measured by latex immunoassay at 6 to 14 and 30 to 36 weeks of gestation, respectively, and the veins of the lower extremities were examined by ultrasound at 30 to 36 weeks of gestation. Results: The mean and standard error of D-dimer was 1.1 ± 1.0 µg/mL in the first trimester and 2.2 ± 1.1 µg/mL in the third trimester, and both values were significantly higher than adult values. In addition, D-dimer significantly increased during pregnancy. D-dimer was not significantly different between singleton and twin pregnancies in the first trimester, but in the third trimester, the values of twin pregnancies were higher than singleton pregnancies (2.2 ± 1.6 vs 3.7 ± 2.5 µg/mL). The mean value of D-dimer of ultrasonographically positive women was 2.6 ± 2.0 µg/mL, which was significantly higher than the value for negative woman during the third trimester (2.2 ± 1.6 µg/mL). The positive predictive value was 7.4% and negative predictive value was 95.5% for ultrasonographically positive women when D-dimer was set at 3.2 µg/mL. Conclusion: We clearly found a change of D-dimer during pregnancy. When D-dimer was higher than 3.2 µg/mL, the percentage of ultrasonographically positive women was high. We propose that women with D-dimer higher than 3.2 µg/mL are closely monitored for prevention of pulmonary thromboembolism. [ABSTRACT FROM AUTHOR]

Edge detection is one of the most basic and important processes in visual signal processing. In order to carry out edge detection from an image corrupted by additive noise, it was necessary to eliminate the noise beforehand by a filter process as a preprocessing. However, since the filter process causes degradation of the original signal itself, the corrupted edge cannot be extracted if the edge detection is carried out after the filter process. The authors have proposed a method for direct edge detection by means of fuzzy inference from the image superposed only with impulsive noise. In this paper, this result is extended to propose a new edge detection method realized by fuzzy inference that carries out both the noise reduction and edge detection at the same time from images contaminated with a mixture of the impulse noise and Gaussian noise. The proposed method consists of two sets of fuzzy inference, one for estimating the number of impulsive noises and another intending to combine the Gaussian noise and edge detection. Finally, by combining these inference results, the edge signal from the mixed noise image is given. It is shown that, by varying the setting of the fuzzy sets for each inference, the degrees of edge detection and noise elimination can be varied easily in a related manner. In addition, the setting of fuzzy sets to satisfy both requirements is carried out by using a typical test image. Further, the effectiveness of the proposed method is shown through various image processing examples. © 2000 Scripta Technica, Electron Comm Jpn Pt 3, 83(8): 39-50, 2000 [ABSTRACT FROM AUTHOR]