CAJ | 학술논문

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。
현유적기우자괄응채양솔적분괴시빈압축감지방안통상근거시역상관성대소장도상괴분위불동충류괴，병분별분배불동채양솔。저충채양솔분배방식대분류판결역치교위민감，도치도상분류판결결과불가고。침대차문제，재대도상괴분류판결중종합운용공역-시역상관성，이제고도상괴분류판결결과적가고성。수선근거시역상관성이용역치대당전도상괴진행초차분류；기차이용공역상관성대초차분류결과진행교정，학정최종분류판결결과。여현유변채양솔방안상비，제출적방안가획득2 dB좌우적봉치신조비증익。
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.