Cairong Zou

A new prediction search algorithm for block motion estimation in video coding

Abstract motion estimation plays an important role for the compression of video signals. This paper presents a new prediction search algorithm (PSA) for block motion estimation. The proposed method utilizes the linear weighting of the motion vectors of the three adjacent blocks to obtain a prediction motion vector, namely, the initial search point. Using a 3/spl times/3 movable search window, it then searches from the prediction point with a fixed search step size of one until it reaches the boundary of search area or its local minimum point lies in the center of search window. Based on the spatial characteristics of the center-biased distance between a motion vector and the corresponding prediction vector and the temporal correlation of motion vectors, this method can significantly reduce the search times. Simulation results show that the proposed PSA is better than the three-step search algorithm and the four-step search algorithm in terms of MSE with smaller computational requirement.

A new algorithm on MPEG-2 target bit-number allocation at scene changes

The bit-number allocation strategy of Test Model 5 (TM5) may cause the coding quality to decrease and the buffer to overflow at scene changes, therefore a new algorithm on MPEG-2 target bit-number allocation is proposed in this letter. Here the influence of scene changes on coding quality is analyzed and an approach to improve the coding performance at scene changes is presented to avoid a sharp quality decrease of scene-change frames. Simulation results indicate that this algorithm provides a promising coding quality at scene changes.

Acoustic feedback cancellation based on weighted adaptive projection subgradient method in hearing aids

Acoustic feedback is an important factor that degrades the overall performance of hearing aids, and acoustic feedback cancellation has always been the research focus in the field of signal processing in hearing aids. The newly suggested adaptive projection subgradient method (APSM) for adaptive signal processing solves the problem of difficulty in finding the exact projection operator in the realization of affine projection by taking the subgradient projection hyperplane as the searching region for relaxed projection. This work applies APSM in the acoustic feedback cancellation system of hearing aids for the first time, and proposes a weighted adaptive projection subgradient method (WAPSM), which takes into consideration the exponential decay weight factor to incorporate the prior information of estimation system. The new method is compared with the traditional NLMS algorithm and APSM algorithm in simulation experiments. Incorporating the prior information of estimation system by setting the proper weighting matrix, WAPSM achieved notable improvements on the speed, stability and accuracy of the misalignment convergence. Numerical experiments demonstrate that the proposed algorithm is more robust for low SNR and real speech segment input than the traditional algorithms.

Multi-level successive elimination algorithm for motion estimation in video coding

In this paper, we present an efficient algorithm to reduce the computation cost of block matching algorithms for motion estimation in video coding. Based on a new insight in block matching algorithms, we extend the successive elimination algorithm to the multi-level case. By using the sum norms of the blocks and the subblocks with reduced sizes, tighter and tighter decision boundaries for eliminating the search positions can be obtained in a multi-level manner. The efficiency of the proposed algorithm combined with full search algorithm and several fast search algorithms is verified by simulation results.

Image sequence macroblock classification using neural networks

Abstract In this paper, a kind of modification to the standard back propagation (SBP) algorithm has been proposed for a real-world macroblock classification problem in video coding. The proposed algorithm studies the effect of adding variance-adaptive noise to the inputs, outputs, and weight connections of multilayer feedforward neural networks during SBP training. Simulations show that the variance-adaptive weight noise is found to be more effective in improving the generalization performance of our problem. Meanwhile adaptive quantization results for moving picture experts group 2 (MPEG-2) video coding confirm the efficacy of the proposed classification method such that it can consistently produce good subjective quality and objective quality sequences at any fixed bit rate. Furthermore, due to the neural networks fault-tolerant nature and inherent parallelism, potential exits for VLSI implementation as a small cell.

An adaptive quantization algorithm for MPEG-2 video coding

An adaptive quantization algorithm for MPEG-2 video coding using a neural network is presented in this paper. The proposed algorithm uses a backpropagation neural network to divide the macroblock activity into one of four categories: flat, edge, texture, fine-texture, and thus the macroblock can be quantized adaptively according to the human vision system (HVS) sensitivity. Experimental results show that this method can reduce block artifacts of flat area and distortion at edge effectively. Meanwhile, the picture subjective quality and objective quality of each frame are improved.

A nonlinear prediction search algorithm with successive differentiation for video coding

A nonlinear prediction search algorithm with successive differentiation (NPSSD) that is used for motion estimation in such international standards as H.261, H.263, MPEG-2, and HDTV is proposed in this paper. The NPSSD method makes full use of the spatial characteristics of center-biased displacement between the motion vector and the corresponding prediction vector. Meanwhile it also adopts the halfway-stop technique and the searching differentiation technique. Therefore this method can significantly reduce the search times. Simulation results show that the proposed NPSSD algorithm is efficient to reduce search times and search complexity of motion estimation. Comparisons of NPSSD with other fast search algorithms are also given in detail.