采用双正交小波和分段正交匹配追踪实现压缩感知图像重构研究

Reconstruction Compressed Sensing Image Renlization Pursuited by Using Biorthogonal Wavelet and Stagewise Orthogonal Matching

采用B样条双正交小波实现图像的稀疏化,用低采样率对高频子带进行压缩感知采样,通过分段正交匹配追踪重建算法实现图像重构.实验结果表明:方案具有较好的图像重构效果,在低通滤波器消失矩相同的情况下,高通滤波器的消失矩越高,图像的压缩感知重构效果越好.随着高通滤波器消失矩的增加,重构图像的效果改善程度减缓,计算复杂性增加.

An image reconstruction is realized by using stagewise orthogonal matching pursuit,B-Spline biorthogonal wavelet for image sparse representation and low sampling frequencies in high frequency subbands for compressed sensing sampling.The experiment results indicate that the better performance on image reconstruction can be achieved.Under the same condition of low-pass filter on vanishing moment,the higher the high-pass filter on vanishing moment is,the better the performance on image reconstruction based on compressed sensing is.Meanwhile,with the increasing of high-pass filter on vanishing moment,the improvement of reconstruction will be slower and the computing complexity will be increased.