中文摘要：

针对传统压缩传感一次性随机测量整幅图像所导致的存储量大、重建时间长等问题，提出了一种新的分块压缩传感重建算法．首先，将图像分割成一系列子块，分别将每个子块的所有列向量首尾连接起来构成原始信号；其次，将该信号经过稀疏变换后投影到观测矩阵上得到对应的观测值，再利用优化方法从这些观测中重建出信号；然后，分类每个重构子块的活动性，采用不同的后滤波方法处理不同活动性指数的子块边界．最后，以MRI图像为实验对象对算法进行了验证．实验结果表明，所提算法不仅降低了压缩传感的重建时间，而且有效去除块效应，并在一定程度上保护图像的纹理和边缘．

英文摘要：

Due to the fact that the traditional compressed sensing （CS） methods need to access the whole image at once in the random sampling process, which results in more storage and reconstruction time, a novel reconstruction algorithm based on block CS is proposed. First, the image is divided into a series of subblocks, and each subblock is concatenated vectors column-wise to represent the original signal. Secondly, the measurement values of the signal are taken by projecting onto a measurement matrix after sparse transformation, and then the recovered signal can be obtained from these measurement values using some optimization methods. Thirdly, through subblock activity classification, the subblock boundaries with different activity indices are processed with different post-filtering methods. Finally, magnetic resonance imaging （MRI） images are used to validate the proposed algorithm. Experimental results show that the proposed approach can decrease reconstruction time and reduce blocking artifacts and preserve textures and edges.