分块子空间追踪算法

Block Subspace Pursuit Algorithm

压缩传感理论是一种充分利用信号稀疏性或者可压缩性的全新信号采样理论。该理论表明,通过采集少量的信号测量值就能够实现可稀疏信号的精确重构。本文在研究现有经典重构算法的基础上,提出结合图像分块思想和回溯思想的分块子空间追踪算法(Block Subspace Pursuit,B_SP)用于压缩传感信号的重构。该算法以块结构获取图像,利用回溯过程实现支撑集的自适应筛选,最终实现图像信号的精确重构。实验结果表明,在相同测试条件下,该算法的重构效果无论从主观视觉上还是客观数据上都有不同程度的提高。

This paper researched the existing classical reconstruction algorithm, and presented a new Block Subspace Pursuit （B_ SP ）algorithm, which combines the blocks thinking and the backing process for reconstruction of signals. The algorithm obtains an image by block structure, using the backing process to achieve the support of set adaptive screening, and ultimately achieves a precise reconstruetion of the image signal. The experimental results show that the proposed algo- rithm can get better reconstruction performances both visually and objectively.