单层小波分解下图像行列压缩感知选择算法

Selection Algorithm of Image Queue Compressed Sensing Based on Single Layer Wavelet Decomposition

中文关键词:  小波分解  压缩感知  图像处理  行列感知 OMP

英文关键词:wavelet decomposition  compressed sensing  image processing  queue sensing  OMP

中文摘要:

      在进行图像压缩感知时发现以行或列进行压缩感知所得到的图像重构后的峰值信噪比(PSNR)是不同的。为了提高图像压缩重构的质量，提出了单层小波分解下图像行列压缩感知的选择算法。该算法首先计算图像的行与列数据的相对方差的最大偏离值，选择较小者对应的行或列作为压缩感知的对象，然后对图像进行单层小波变换分解出高频系数，在高斯观测矩阵下，对这些系数按指定的行或列进行压缩感知，最后利用正交匹配追踪算法(OMP)分别恢复压缩感知下的高频系数，并通过小波逆变换得到经过行列压缩感知后的重构图像，实验结果证明了算法的准确性。

英文摘要:

      It is found that the peak signal-to-noise ratio (PSNR) of reconstructed images is different from the row or column of compressed sensing.In order to improve the quality of image compression and reconstruction,a selection algorithm of image queue compressed sensing based on single layer wavelet decomposition is proposed.The algorithm first calculates the maximum deviation of the relative variance of the image rows and columns data,selects the smaller corresponding row or column as the object of compressed sensing,and then uses single layer wavelet transform to decompose the high frequency coefficients,and under the Gauss observation matrix,the coefficients are compressed sensing by the specified row or column.Finally,the orthogonal matching pursuit algorithm (OMP) is used to recover the high frequency coefficients respectively,and reconstruct the image by the inverse wavelet transform under row or column compression sensing.The experimental results verify the accuracy of the algorithm.