基于LDPC测量矩阵的压缩感知图像重建

Compressed Sensing Image Reconstruction Based on LDPC Measurement Matrix

测量矩阵是影响压缩感知信号重构效果的关键因素。为了研究重构性能好的测量矩阵,提出构造基于低密度奇偶校验码(LDPC)正则校验矩阵的测量矩阵。首先从理论上分析LDPC校验矩阵的性质,验证其有效性,然后采用渐进边增长算法构造矩阵,最后实验选取3个自然图像进行压缩采样重建,与常规的3种随机测量矩阵进行对比。实验结果表明,提出的测量矩阵重构性能良好,能提高压缩图像的分辨率、重建峰值信噪比,降低均方误差。

Measurement matrix is the key factor that affects the reconstruction effect of compressed sensing signal.In order to study the measurement matrix with good reconstruction performance,a measurement matrix based on LDPC regular check matrix is proposed.Firstly,the properties of the LDPC check matrix are analyzed theoretically to verify its effectiveness,and then the matrix is constructed by using the progressive edge growth algorithm.Finally,three natural images are selected for compression sampling and reconstruction,and compared with the conventional three random measurement matrices.The experimental results show that the proposed measurement matrix reconstruction has good performance which can improve the resolution of compressed image and reconstruct the peak signal to noise ratio,and reduce the mean square error.