基于空时相关性的变采样率分块视频压缩感知

Adaptive Block Compressive Video Sensing Scheme Based on Spatial-Temporal Correlation

现有的基于自适应采样率的分块视频压缩感知方案通常根据时域相关性大小将图像块分为不同种类块,并分别分配不同采样率。这种采样率分配方式对分类判决阈值较为敏感,导致图像分类判决结果不可靠。针对此问题,在对图像块分类判决中综合运用空域-时域相关性,以提高图像块分类判决结果的可靠性。首先根据时域相关性利用阈值对当前图像块进行初次分类;其次利用空域相关性对初次分类结果进行校正,确定最终分类判决结果。与现有变采样率方案相比,提出的方案可获得2 dB左右的峰值信噪比增益。

For most existing block compressed video sensing schemes with adaptive sampling rate, image blocks are divided into different types of block with different allocated sampling rate according to temporal correlation. However, the classification procedure is sensitive to the judgment threshold, which leads to unreliable result. In order to solve this problem, spatial correlation and temporal correlation are jointly considered in this paper to improve the reliability of the block classification procedure. Fist, image blocks are divided into different types according to threshold. Then the initial classification is corrected accroding to spatial correlation to determin final classification decision. Compared with the existing algorithms, the proposed one can get about 2 dB Peak Signal to Noise Ratio（PSNR） incerement at the same sampling rate.