Search

Dictionary learning has achieved remarkable success on a wide range of machine learning-based applications. In this paper, a locality-adaptive structured dictionary learning (LASDL) algorithm for cross-domain recognition is proposed. In the LASDL, a projective structured double reconstruction strategy is developed to train class-oriented sub-dictionaries from the specific classes of cross-domain samples. The strategy benefits to make full advantage of the discriminative information of cross-domain data and bridge the distribution divergence between two different domains. Meanwhile, an adaptive geometrical structure preserving function is designed to not only preserve the local manifold structures spanned by the representation coefficient spaces of the source and target domains, but also impose a constraint that the coefficients should keep closer to their class centers, which is propitious to reduce the distribution divergence and make the representation more accurate. With the structured linear coding technique, the final cross-domain recognition can be efficiently performed by determining class-specific reconstruction error. The optimization of the proposed LASDL model can be efficiently solved by simple least square method and alternating direction method of multipliers (ADMM) algorithm. Extensive experimental results validate the superiority of the proposed algorithm in contrast to other state-of-the-art predecessors. [ABSTRACT FROM AUTHOR]

As a specific data hiding technique, reversible data hiding (RDH) has recently received extensive attention. By this technique, both the embedded data and the original cover image can be exactly extracted from the marked image. In our previous work, a general expansion-shifting model for RDH is proposed by introducing the so-called reversible embedding function (REF). With REF, RDH can be designed and the corresponding rate-distortion formulations can be established, providing an approach to optimize the reversible embedding performance. In this paper, by extending our previous work, optimal REF for one-dimensional histogram is investigated, and all optimal REF are derived in this case when the maximum modification to the cover pixel is limited as a small value. Moreover, based on the derived optimal REF for one-dimensional histogram and multiple histograms modification, a practical RDH scheme is presented and it is experimental verified better than some state-of-the-art algorithms in terms of capacity-distortion performance. [ABSTRACT FROM AUTHOR]

Considering that traditional image steganography technologies suffer from the potential risk of failure under lossy channels, an enhanced adaptive steganography with multiple robustness against image processing attacks is proposed, while maintaining good detection resistance. First, a robust domain constructing method is proposed utilizing robust element extraction and optimal element modification, which can be applied to both spatial and JPEG images. Then, a robust steganography is proposed based on “Robust Domain Constructing + RS-STC Codes,” combined with cover selection, robust cover extraction, message coding, and embedding with minimized costs. In addition, to provide a theoretical basis for message extraction integrity, the fault tolerance of the proposed algorithm is deduced using error model based on burst errors and decoding damage. Finally, on the basis of parameter discussion about robust domain construction, performance experiments are conducted, and the recommended coding parameters are given for lossy channels with different attacks using the analytic results for fault tolerance. A series of experimental results demonstrate that the proposed algorithm can extract embedded messages with significantly higher accuracy after different attacks, such as compression, noising, scaling and other attacks, compared with the state-of-the-art adaptive steganography, and robust watermarking algorithms, while maintaining good detection resistant performance. [ABSTRACT FROM AUTHOR]

Among various methods of reversible data hiding (RDH) in JPEG images, only rate-distortion, i.e. the image quality with given payload, is taken into consideration during algorithm designing. However, file size expansion is another important evaluation metric for JPEG RDH methods. Based on this situation, we propose a JPEG RDH method considering both the rate-distortion and the file size expansion at the same time while designing the algorithm. The multi-objective optimization strategy is utilized to realize the balance of the two objectives. Specifically, the cover signal is divided into several non-overlapping parts firstly, and after that, the embedding costs of each part are calculated. Next, the optimized combination of parts for embedding data is gained by multi-objective optimization. Experimental results show that the proposed algorithm outperforms the state-of-the-art methods in terms of rate-distortion and file size expansion performance. [ABSTRACT FROM AUTHOR]

Low-dose computed tomography (LDCT) images are often highly degraded by amplified mottle noise and streak artifacts. Maintaining image quality under low-dose scan protocols is a well-known challenge. Recently, sparse representation-based techniques have been shown to be efficient in improving such CT images. In this paper, we propose a 3D feature constrained reconstruction (3D-FCR) algorithm for LDCT image reconstruction. The feature information used in the 3D-FCR algorithm relies on a 3D feature dictionary constructed from available high quality standard-dose CT sample. The CT voxels and the sparse coefficients are sequentially updated using an alternating minimization scheme. The performance of the 3D-FCR algorithm was assessed through experiments conducted on phantom simulation data and clinical data. A comparison with previously reported solutions was also performed. Qualitative and quantitative results show that the proposed method can lead to a promising improvement of LDCT image quality. [ABSTRACT FROM PUBLISHER]

In this paper, we propose an efficient blind image quality assessment (BIQA) algorithm, which is characterized by a new feature fusion scheme and a k -nearest-neighbor (KNN)-based quality prediction model. Our goal is to predict the perceptual quality of an image without any prior information of its reference image and distortion type. Since the reference image is inaccessible in many applications, the BIQA is quite desirable in this context. In our method, a new feature fusion scheme is first introduced by combining an image’s statistical information from multiple domains (i.e., discrete cosine transform, wavelet, and spatial domains) and multiple color channels (i.e., Y, Cb, and Cr). Then, the predicted image quality is generated from a nonparametric model, which is referred to as the label transfer (LT). Based on the assumption that similar images share similar perceptual qualities, we implement the LT with an image retrieval procedure, where a query image’s KNNs are searched for from some annotated images. The weighted average of the KNN labels (e.g., difference mean opinion score or mean opinion score) is used as the predicted quality score. The proposed method is straightforward and computationally appealing. Experimental results on three publicly available databases (i.e., LIVE II, TID2008, and CSIQ) show that the proposed method is highly consistent with human perception and outperforms many representative BIQA metrics. [ABSTRACT FROM AUTHOR]

Bit rate is one of the important criterions for digital video quality. With some video tools, however, video bit rate can be easily increased without improving the video quality at all. In such a case, a claimed high bit rate video would actually have poor visual quality if it is up-converted from an original lower bit rate version. Therefore, exposing fake bit rate videos becomes an important issue for digital video forensics. To the best of our knowledge, although some methods have been proposed for exposing fake bit rate MPEG-2 videos, no relative work has been reported to further estimate their original bit rates. In this paper, we first analyze the statistical artifacts of these fake bit rate videos, including the requantization artifacts based on the first-digit law in the DCT frequency domain (12-D) and the changes of the structural similarity indexes between the query video and its sequential bit rate down-converted versions in the spatial domain (4-D), and then we propose a compact yet very effective 16-D feature vector for exposing fake bit rate videos and further estimating their original bit rates. The extensive experiments evaluated on hundreds of video sequences with four different resolutions and two typical compression schemes (i.e., MPEG-2 and H.264/AVC) have shown the effectiveness of the proposed method compared with the existing relative ones. [ABSTRACT FROM AUTHOR]

A low-cost very large scale integration (VLSI) implementation of real-time correction of barrel distortion for video-endoscopic images is presented in this paper. The correcting mathematical model is based on least-squares estimation. To decrease the computing complexity, we use an odd-order polynomial to approximate the back-mapping expansion polynomial. By algebraic transformation, the approximated polynomial becomes a monomial form which can be solved by Hornor's algorithm. With the iterative characteristic of Hornor's algorithm, the hardware cost and memory requirement can be conserved by time multiplexed design. In addition, a simplified architecture of the linear interpolation is used to reduce more computing resource and silicon area. The VLSI architecture of this paper contains 13.9-K gates by using a 0.18 \mum CMOS process. Compared with some existing distortion correction techniques, this paper reduces at least 69% hardware cost and 75% memory requirement. [ABSTRACT FROM AUTHOR]

A novel scaling algorithm is proposed for the implementation of 2-D image scalar. The algorithm consists of a bilinear interpolation, a clamp filter, and a sharpening spatial filter. The bilinear interpolation algorithm is selected due to its having low complexity and high quality. The clamp and sharpening spatial filters are added as pre-filters to solve the blurring and aliasing effects produced by bilinear interpolation. Furthermore, an adaptive technology is used to enhance the effects of clamp and sharpening spatial filters. To reduce memory buffers and computing resources for the very large scale integration (VLSI) implementation, the clamp filter and sharpening spatial filters both convoluted by a 3\,\times\,3 matrix coefficient kernel are combined into a 5\,\times\,5 combined convolution filter. The bilinear interpolation is simplified by the co-operation and hardware sharing technique to reduce computing resource and hardware costs. The VLSI architecture in this paper can achieve 280 MHz with 9.28-K gate counts, and its chip area is 46 418 \mum^2 synthesized by a 0.13 \mum CMOS process. Compared with previous techniques, this paper not only reduces gate counts by more than 46.6% and power consumptions by 24.2%, but also improves average quality by over 0.42 dB. [ABSTRACT FROM AUTHOR]

Recurring events are short temporal patterns that consist of multiple instances in the target database. Without any a priori knowledge of the recurring events, in terms of their lengths, temporal locations, the total number of such events, and possible variations, it is a challenging problem to discover them because of the enormous computational cost involved in analyzing huge databases and the difficulty in accommodating all the possible variations without even knowing the target. We translate the recurring event mining problem into finding temporally continuous paths in a matching.trellis. A novel algorithm that simulates a "forest-growing" procedure in the matching-trellis is proposed. Each tree branch in the resulting forest naturally corresponds to a discovered repetition, with temporal and content variations tolerated. By using locality sensitive hashing (LSH) to find best matches efficiently, the overall complexity of our algorithm is only sub-quadratic to the size of the database. Experimental results on the TRECVID video data of 10.5 hours and a human dance video dataset of 32,260 frames show that our method can effectively and efficiently discover recurring events such as TV commercials from news videos and typical dance moves from human dance sequences, in spite of large temporal and content variations. [ABSTRACT FROM AUTHOR]

Motion estimation is a multistep process that encompasses techniques such as motion vector prediction, determination of search range and search patterns, and identification of termination criteria. Each of these techniques has several diversions that may suit a particular set of video characteristics. It would be hard to conceive a universal algorithm that can perform well for all kinds of video contents. However, if important characteristics of a video sequence can be identified and utilized for adjusting various steps of motion estimation, one can design an adjustable algorithm that can tune its parameters to suit the video at hand. A multistage motion estimation algorithm that includes a pre-stage for analyzing the motion characteristics of a video sequence online is proposed. This stage predicts the motion vector field (MVF) from the previous coded frame, and clusters macroblocks into background and foreground regions based on the predicted MVF. The information from the pre-stage are passed on to the next stage, which includes a mathematical model for block distortion surface to estimate the distance from the current search point to the global optimal position. This allows the motion estimation stage to adjust its search strategies accordingly. The search is performed on a precise area according to the statistical properties of the motion vector prediction error, separately for foreground and background regions. The proposed algorithm, including the pre-stage, is fast and, therefore, is suitable for online and real-time encoding. Extensive simulation results obtained for several video sequences affirm the effectiveness of the proposed algorithm. The self-tuning property enables the algorithm to perform well for several types of benchmark sequences, yielding better video quality without exceeding the time complexity as compared to the other predictive motion estimation algorithms. [ABSTRACT FROM AUTHOR]

Motion estimation (ME) is a multistep process that involves not one, but a combination of techniques, such as motion starting point, motion search patterns, and adaptive control to curb the search, avoidance of search stationary regions, etc. The collective efficiency of these techniques is what makes a ME algorithm robust and efficient across the board. This paper proposes a ME algorithm that is an embodiment of several effective ideas for finding the most accurate motion vectors (MVs) with the aim to maximize the encoding speed as well as the visual quality. The proposed algorithm takes advantage of the correlation between MVs in both spatial and temporal domains, controls to curb the search, avoids of search stationary regions, and uses switchable shape search pattern to accelerate motion search. The algorithm yields very similar quality compared to the full search but with several hundred times faster speed. We have evaluated the algorithm through a comprehensive performance study that shows that the proposed algorithm achieves substantial speedup without quality loss for a wide range of video sequences, compared with the ME techniques recommended by the MPEG-4 committee. [ABSTRACT FROM AUTHOR]

This paper presents a game theory based technique for optimizing the bit rate control in video coding. Game theory, by virtue of its enormous potential for solving constrained optimization problems, has been effectively utilized in several branches of natural and social sciences. But this paper is the first attempt in using game theory for video compression. The objective is to optimize the perceptual quality while guaranteeing "fairness" in bit allocation among macroblocks (MBs). The proposed technique is a dual-level rate control algorithm: At the first level, the algorithm allocates the target bits to frames based on their coding complexity; a method to estimate the coding complexity of the remaining frames is proposed. At the second level, MBs of a frame play cooperative games such that each MB competes for a fair share of resources (bits) to optimize its quantization scale while considering the human visual system (HVS) perceptual property. We formulate the rate control problem by defining players, strategies and objective function. Since the whole frame is an entity perceived by viewers, MBs compete cooperatively under a global objective of achieving the best quality with the given bit constraint. The major advantage of the proposed approach is that the cooperative game leads to an optimal and fair bit allocation strategy based on the Nash bargaining solution. Another advantage is that it allows multi-objective optimization with multiple decision makers (e.g., MBs) in order to achieve accurate bit rate with good perceptual quality while maintaining a stable buffer level. Several extensions of the work are possible. [ABSTRACT FROM AUTHOR]

The embedded multireference frames selection with variable block-size motion compensation model drastically in- creases the computational complexity of the H.264/AVC video coding standard. This paper proposes an adaptive hierarchical motion estimation (ME) algorithm for H.264/AVC with the objective of minimizing the complexity while maximizing the visual quality. The proposed algorithm is based on a framework that exploits the "history" of the motion intensity from a video sequence in order to control ME. The complexity and memory requirement for this meta information is low. The algorithm determines the motion intensity of a video sequence at three levels and accordingly employs different ME techniques. The results certify that the history-based hierarchical information can be very effective in improving the efficiency of ME. [ABSTRACT FROM AUTHOR]

We present a new shape-coding algorithm to support object-based representation, which differs from previous algorithms in that it encodes shape as dependent meta data for image description. Therefore, both the shape-coding and decoding processes of this algorithm are designed to be dependent on the underlying image in which the object (described by the shape) is contained. This way, the correlation between image and shape is effectively removed and the shape-coding efficiency is improved on average by three times over the state-of-the-art algorithms. To facilitate comparison, a generalized "contour-generating" frame-work is introduced to formulate the shape-coding problem. From this framework we derive both the proposed algorithm and a number of state-of-the-art algorithms, and show that the rate-distortion (RO) criterion can be studied in a uniform way under this framework. Specifically, a dynamic-programming-based algorithm is designed to find the RD optimal coding result for the proposed algorithm. As an extension, we also discuss the complexity and scalability issues related to the application design of the proposed algorithm. [ABSTRACT FROM AUTHOR]

Presents a fast motion estimation algorithm using binary representation for embedded system and hardware implementation with parallel architectures. Key algorithm distinction; Analysis of specific implementation techniques for multimedia-embedded system design; Use of various filters and binarization methods.