Search

Reflection caused by glass often degrades the quality of an image and further makes it difficult to estimate depth. In this article, we propose joint reflection removal and depth estimation from a single image. We perform reflection removal (transmission recovery) and depth estimation jointly using a collaborative neural network that consists of four blocks: 1) encoder for feature extraction; 2) reflection removal subnetwork (RRN); 3) depth estimation subnetwork (DEN); and 4) depth refinement guided by the transmission layer. We achieve collaboration between reflection removal and depth estimation by concatenating intermediate features of DEN with RRN. Since the recovered transmission layer contains accurate edges of objects behind glass, we refine the estimated depth with its guidance by guided image filtering. The experimental results demonstrate that the proposed method achieves both reflection removal and depth estimation even for images with dominant reflections. Besides, this article offers a new way of treating reflections in images to introduce depth estimation into reflection removal and achieve reflection removal and depth estimation simultaneously. [ABSTRACT FROM AUTHOR]

Remote sensing images obtained from onboard linear array cameras suffer from geometric disturbances in the presence of attitude jitter from a satellite platform. Thus, platform jitter estimation is essential for improving satellite stability and enhancing remote sensing image quality. In this article, a novel integration framework is designed to estimate attitude fluctuation, which combines multispectral advanced spaceborne thermal emission and reflection/short-wave-infrared (ASTER/SWIR) data and blurred star images from onboard optical sensors. First, a multilevel XGBoost algorithm is applied to learn the nonlinear relationship between the star blurring and instantaneous jitter displacement. An image registration method and a deconvolutional process are then introduced to remove the satellite jitter from ASTER/SWIR data. Next, an augmented $H$$_\infty$ filter is proposed to fuse and estimate the satellite jitter from two algorithms. Simulating experiment results indicate that the jitter estimation error of the proposed integration framework is reduced by 40%. Compared with the existing jitter estimation methods only based on remote sensing image processing, our strategy has better robustness and accuracy, especially on extreme ground scenes. [ABSTRACT FROM AUTHOR]

This article proposes a new position estimation method based on an unsaturated flux linkage model for planar switched reluctance motors (PSRMs). The bipolar-voltage-injection method is used to measure the flux linkage, and it is found that ripples caused by iron loss and aluminum eddy current with different amplitudes appear at the troughs of the flux curve. The effect of the losses on the flux linkage measurement is revealed in this article, and the loss compensation models are built based on the analysis. The improved flux linkage model is proposed with the loss compensation models. Then, the position estimation method is given with the improved flux linkage model. Flux linkage measurement result shows that the model with loss compensation has higher accuracy than the conventional model. Initial and planar-motion position estimation experiments are carried out for the PSRM. The initial position estimation error is from −0.4 to 0.4 mm. The planar-motion position estimation errors are from −1.1 to 1.1 mm in the ${X}$ -axis and from −1.2 to 1.0 mm in the ${Y}$ -axis, and the effectiveness of the proposed flux linkage model is verified. [ABSTRACT FROM AUTHOR]

In this article, a position estimation method is proposed to realize sensorless control of planar switched reluctance motor (PSRM) with core loss. First, a square-wave voltage generated through bipolar pulsewidth modulation (PWM) is injected into the idle phase of a PSRM, thereby generating core loss. The corresponding core-loss average power (CLAP), which contains position information, is calculated in real time. Then, depending on the six-phase CLAP, the initial position can be estimated for the PSRM startup. To stabilize the position estimation in different regions of pole pitches, position data measured in a long stroke are utilized to determine a function mapping from the CLAP with respect to position. Furthermore, a sensorless control system and commutation strategy are designed. Finally, the capability of different methods for position estimation and sensorless control of a PSRM system are experimentally investigated. The feasibility and effectiveness of the proposed method are verified through the experimental results. [ABSTRACT FROM AUTHOR]

An accurate mover position is important for a planar switched reluctance motor (PSRM) to achieve precision positioning; however, position sensors are susceptible to the influence from harsh conditions and are relatively expensive. In this article, an assistant-mover-based position estimation method is proposed to increase the reliability and reduce the cost of a PSRM system. The method is based on a mapping relationship between the flux density and position of the assistant mover. First, the configuration of the assistant movers is presented. Then, an analytical model of the flux density of the assistant mover is developed by using an equivalent magnetic circuit method. Based on the analytical model, the parameters of the assistant mover are determined. According to the flux density characteristics, a position estimation algorithm with the advantages of not requiring previous experimental data, simple computation, and small memory consumption is developed. Moreover, considering the deviation in the characteristics of each assistant mover, a compensation strategy is proposed to obtain better position estimation accuracy. Finally, the proposed method is implemented with the PSRM system. The effectiveness of the proposed position estimation method is verified experimentally. [ABSTRACT FROM AUTHOR]

Foreground detection is an important task in computer vision applications. In this paper, we present an efficient foreground detection method based on a robust linear regression model. First, a novel framework is proposed where foreground detection has been cast as an outlier signal estimation problem in a linear regression model. We regularize this problem by imposing a so-called fused sparsity constraint, which encourages both sparsity and smoothness of vector coefficients, on the outlier signal. Second, we convert this outlier signal estimation problem into an equivalent Fused Lasso problem, and then use existing solutions to obtain an optimized solution. Third, a new foreground detection method is presented to apply this new model to the 2-D image domain by merging the results from different vectorizations. Experiments on a set of challenging sequences show that the proposed method is not only superior to many state-of-the-art techniques, but also robust to noise. [ABSTRACT FROM PUBLISHER]

In orthogonal frequency division multiple access (OFDMA) uplink system, estimations of time delay and carrier frequency offset (CFO) are both particularly challenging. In IEEE 802.16-based networks, such synchronization problem can be resolved in initial ranging process. In this paper, exploiting the unique scheme of interleaved OFDMA, a special ranging symbol which has repetitive scheme in time-domain is designed. Furthermore, a novel ranging method for OFDMA system is proposed by utilizing such repetitive characteristic. Compared with the existing methods, this algorithm just need one ranging symbol and can get more correct estimation of time delay, at the same time, it can estimate the integral CFO in OFDMA uplink. Simulations illustrate efficiency of the proposed algorithm over the existing method. [ABSTRACT FROM PUBLISHER]