中山大学学报(自然科学版)
中山大學學報(自然科學版)
중산대학학보(자연과학판)
ACTA SCIENTIARUM NATURALIUM UNIVERSITATIS SUNYATSENI
2013年
3期
30-33
,共4页
多尺度%DWT%压缩感知 (CS)%变采样率%重构
多呎度%DWT%壓縮感知 (CS)%變採樣率%重構
다척도%DWT%압축감지 (CS)%변채양솔%중구
multiscale%DWT%compressed sensing%variable sampling rate%reconstruction
利用压缩感知理论改善图像重构的质量是目前图像处理技术研究的焦点。通过DWT域对图像每级分解时的每个子带中应用分块采样并结合平滑投影Landweber重构算法,提出一种多尺度分块变采样率压缩感知图像重构算法。比较BCS-SPL 和TV 以及多尺度GPSR图像处理算法,文中提出的算法使重构的图像质量提高了1~3 dB。
利用壓縮感知理論改善圖像重構的質量是目前圖像處理技術研究的焦點。通過DWT域對圖像每級分解時的每箇子帶中應用分塊採樣併結閤平滑投影Landweber重構算法,提齣一種多呎度分塊變採樣率壓縮感知圖像重構算法。比較BCS-SPL 和TV 以及多呎度GPSR圖像處理算法,文中提齣的算法使重構的圖像質量提高瞭1~3 dB。
이용압축감지이론개선도상중구적질량시목전도상처리기술연구적초점。통과DWT역대도상매급분해시적매개자대중응용분괴채양병결합평활투영Landweber중구산법,제출일충다척도분괴변채양솔압축감지도상중구산법。비교BCS-SPL 화TV 이급다척도GPSR도상처리산법,문중제출적산법사중구적도상질량제고료1~3 dB。
At present, utilizing Compressed Sensing (CS) theory to improve image quality of recon-structions is a research focus of image process technology .By deploying block-based CS variable sampling within the domain of a wavelet transform , a multiscale block-CS algorithm with smoothed projected Land-weber reconstruction algorithm is provided .Block-based CS sampling is applied independently within each subband of each decomposition level of a wavelet transform of a image .Experimental results reveal that the algorithm achieves a 1~3 dB gain in reconstruction PSNR over the BCS-SPL and TV as well as multiscale GPSR algorithm.