Chengke Wu

Image dense matching based on region growth with adaptive window

Abstract A new and efficient dense matching algorithm is presented based on region growth, which can be applied to a wide range of image pairs including those with large disparity or without rectification. Firstly, some points in the image pair are matched using a new two-level matching method. These points are taken as the seeds from which corresponding relations propagate towards other regions of the images under two strategies. By introducing the statistic model of disparity distribution within the window, modified SSD is produced and adopted as the cost function. The size of the template window is adaptive with textures within it and the size of the search window is changed in inverse proportion to the confidence coefficient. The algorithm has been tested with real stereo images and the results demonstrate its accuracy and efficiency.

Performance analysis of flexible hierarchical cellular systems with a bandwidth-efficient handoff scheme

In this paper, a bandwidth-efficient handoff strategy is proposed and analyzed for hierarchical cellular systems. Mobile subscribers are divided into two classes, i.e., low- and high-mobility subscribers. In our bandwidth-efficient handoff strategy, each of the originating and handoff calls of both slow and fast mobile subscribers first tries to get a channel in a microcell. Macrocells are overlaid over the microcells to handle overflowed calls. A call overflowed into a macrocell requests a take-back to the new microcell at each border crossing of the microcells. The request will be accommodated by the target microcell if there is any idle traffic channel in the cell. An analytical model is developed, and the most important performance measures such as the blocking probability of originating calls and the forced termination probability of calls are evaluated. It is shown through extensive comparisons with other candidate handoff strategies that if the total traffic load of the system is not very heavy, our scheme has the best bandwidth efficiency and can provide better quality of service for mobile subscribers without bringing too much processing expense to the system.

A new image rectification algorithm

A novel and efficient image rectification method using the fundamental matrix is proposed. In this approach, camera calibration is not required, and image resampling becomes very simple by using the Bresenham algorithm to extract pixels along the corresponding epipolar line. The rectified images are guaranteed to be effective for all possible camera motions, large or small. The loss of pixel information along the epipolar lines is minimized, and the size of rectified image is much smaller. Furthermore, it never splits the image and the connected regions will stay connected, even if the epipole locates inside an image. The effectiveness of our method is verified by an extensive set of real experiments. It shows that much more accurate matches of feature points can be obtained for a pair of images after the proposed rectification.

Lossy to lossless compressions of hyperspectral images using three-dimensional set partitioning algorithm

In this paper, we present a three-dimensional (3D) hyperspectral image compression algorithm based on zeroblock coding and wavelet transforms. An efficient Asymmetric 3D wavelet Transform (AT) based on the lifting technique and packet transform is used to reduce redundancies in both the spectral and spatial dimensions. The implementation via 3D integer lifting scheme allows to map integer-to-integer values, enabling lossy and lossless decompression from the same bit stream. To encode these coefficients after Asymmetric 3D wavelet transform, a modified 3DSPECK algorithm - Asymmetric Transform 3D Set Partitioning Embedded bloCK (AT-3DSPECK) is proposed. According to the distribution of energy of the transformed coefficients, the 3DSPECKs 3D set partitioning block algorithm and the 3D octave band partitioning scheme are efficiently combined in the proposed AT-3DSPECK algorithm. Several AVIRIS images are used to evaluate the compression performance. Compared with the JPEG2000, AT-3DSPIHT and 3DSPECK lossless compression techniques, the AT-3DSPECK achieves the best lossless performance. In lossy mode, the AT-3DSPECK algorithm outperforms AT-3DSPIHT and 3DSPECK at all rates. Besides the high compression performance, AT-3DSPECK supports progressive transmission. Clearly, the proposed AT-3DSPECK algorithm is a better candidate than several conventional methods.

A robust algorithm to estimate the fundamental matrix

Abstract A new method, the biepipole constraint algorithm, is developed to estimate the fundamental matrix (F-matrix) based on an 8-parameter model and the geometrical analysis. First, through the analysis of the new constraints, the four parameters of the F-matrix can be estimated by solving a nonlinear unconstraint optimization problem. The objective function of the optimization problem is an equation of degree six in four unknowns. Then, the four other parameters of the F-matrix can be evaluated by using the SVD method. Particular novelties of the algorithm are the obvious geometrical meanings of the parameters, fewer matching point pairs and higher accuracy. Results are presented for synthetic and real image pairs, which show that our algorithm performs very well in terms of robustness to outliers and noises, so that excellent results can be obtained.

Wyner-Ziv video coding with block classification

Wyner-Ziv (WZ) video coding, or lossy coding of video with decoder side information (SI), is a new video coding paradigm which provides the potential of shifting encoder complexity to decoder. Due to the fact that correlation among adjacent frames varies spatially and temporally in a video sequence, in WZ video coding, the quality of SI generated at the decoder also varies accordingly. This implies that more parity bits should be sent for locations where the quality of the SI is poor, while fewer parity bits for locations where the quality of the SI is good. For this reason, this paper proposes a WZ video coding scheme with block classification. In the scheme, according to difference between blocks in a WZ frame and its adjacent key frame, the blocks in a WZ frame are classified into classes of different importance. A strategy is designed so that the more important a class is, the more bits will be transmitted for it, and this leads to savings in bit rate. Simulation on various test video sequences shows that, compared to a similar scheme without block classification, the proposed scheme can achieve a bit rate savings of up to 42.44% with only slight PSNR loss in some cases.

A new fast motion estimation algorithm based on the loop-epipolar constraint for multiview video coding

There lie geometric constraints between neighboring frames in multiview video sequences. The geometric constraints are valuable for reducing spatial and temporal redundancy in multiview video coding (MVC). In this paper, we propose a new fast motion estimation algorithm based on the loop-epipolar constraint which combines loop and epipolar constraints. A practical search technique is designed according to the characteristics of the loop-epipolar constraint. Experimental results show that the proposed algorithm is efficient for sequences under different multiview camera setups. Highlights? A fast motion estimation algorithm is proposed based on the loop-epipolar constraint. ? The loop-epipolar constraint combines loop and epipolar constraints. ? A practical search technique is designed according to the characteristics of the loop-epipolar constraint. ? The proposed algorithm is efficient for different sequences.

A novel homography-based search algorithm for block motion estimation in video coding

There lies a geometric relationship (homography) between two frames in the video sequences captured by pan-tilt (PT) cameras due to their constrained movement, and the geometric relationship is valuable for reducing the spatial redundancy in video coding. In this paper, we propose a novel homography-based search (HBS) algorithm for block motion estimation in coding the sequences captured by PT cameras, which well utilizes the homography between two frames. In addition, adaptive thresholds are adopted in our method to classify different kinds of blocks. Compared with other traditional fast algorithms, the proposed HBS algorithm is proved more efficient for the sequences captured by PT cameras.

Divide sampling–based hybrid temporal-spatial prediction coding for H.264/AVC

A divide sampling-based hybrid temporal-spatial prediction coding algorithm is designed to further improve the coding performance of the conventional H.264/AVC coding. In the proposed algorithm, a frame is first divided into four equal-sized subframes, and the first subframe is coded using the rate distortion optimization model with inter- or intraprediction adaptively. Then, the optimal prediction method of the macroblock in other subframes is selected flexibly and reasonably from intraprediction, the fast interprediction, and the spatial interpolation prediction. The simulation results show that compared with the conventional H.264/AVC coding, the average bit rate is reduced by 6.15% under the same peak signal-to-noise ratio (PSNR), the average PSNR is increased by 0.22 dB under the same bit rate, and the average coding time is saved by 12.40% in the proposed algorithm.

An Approach to Side Information Estimation for Wyner-Ziv Video Coding

In the new video coding paradigm, Wyner-Ziv (WZ) video coding, or lossy coding of video with decoder side information (SI), SI is involved not only in the decoding of the quantized source, but also in the reconstruction process, so the quality of SI is a key factor to the performance of a WZ video codec. In this paper, an approach to SI estimation for WZ video coding is proposed. It is based on motion compensated temporal interpolation. In the approach, stationary areas are first identified by motion detection. Nearly textureless areas for which motion vectors can not be reliably estimated with smaller block size are also identified by the analysis of spatial intensity gradients. Based on the identification of such areas and on the matching errors of the estimated motion vectors, a variable size block based motion field is then estimated using a splitting strategy. Post-processing of the estimated motion vectors is also conducted to deal with blocks which may still contain more than one objects with different motion. The performance of the proposed approach is evaluated in the context of WZ video coding, and compared with some existing methods. The simulation results verify the effectiveness of the propose approach.