Use the compressed sensing theory to realize the accurate reconstruction of two-dimensional image,but this method aims at whole image which is needed long time and large storage space of matrix. To solve this problem,according to the properties of image wavelet transform coefficients,combined image blocking theory and wavelet transform,an improved blocking compressed sensing algo-rithm based on wavelet transform is proposed,which only measured the high-pass wavelet coefficients of the image algorithm while retai-ning the low-dimensional coefficients. For the reconstruction,by using the Orthogonal Matching Pursuit ( OMP) algorithm,high-pass wavelet coefficients could be recovered by the measurements. Experimental results show that the PSNR of image reconstruction is im-proved about 2 to 4,the time of reconstruction is decreased obviously,compared with the compressed sensing algorithm without blocking. Compared with blocking compressed sensing based on DCT,the blocking effects and the quality of the recovered image is improved obvi-ously.%利用压缩感知理论实现了对二维图像的精确重构，但此方法是对整幅图像进行重构，花费时间长，观测矩阵所需的存储空间大。为了解决这个矛盾，根据图像小波变换系数的特点，将图像分块思想与小波变换相结合，提出一种基于小波变换的分块压缩感知算法。每一个图像块经小波变换后，保留图像低频系数，只对高频系数进行观测。重构时采用正交匹配追踪算法( OMP)对高频系数进行恢复。实验结果表明，文中算法与不分块压缩感知算法相比，重构图像的PSNR值有2~4 dB的提高，重构时间明显减少。与基于二维离散余弦变换( DCT)的分块压缩感知算法相比，块效应有明显的改善，重构质量明显提高。
展开▼
