基于快速基追踪算法的图像去噪

基追踪方法是信号稀疏表示领域的一种新方法。基于基追踪方法的核心思想，并对基追踪求解算法进行改进，本文将基追踪方法的应用扩展到图像去噪问题上来。〖BP）〗针对基追踪方法中线性规划算法计算量大的难题，考虑了一类可分离的信号表示形式，分别在图像的行和列的方向上构造相应的字典，由此降低了字典的维数，并将一个二维问题转化为两个一维问题来处理，最后利用一种新的迭代算法进行求解，从而加快了计算速度。实验结果表明，改进的基追踪方法能够快速稳定实现，同时具有较好的去噪效果。

Image de-noising based on fast basis pursuit algorithm

Aiming at the hardship to solve one kind large-scale linear programming problem of Basis Pursuit, certain kind form of separable signal representation was considered. We construct dictionaries in column-wise and row-wise of the image, the dimension of the dictionary was greatly reduced. Then two-dimension problem was turning into two one - dimension problem. Finally, one new iterative method was presented to solve this problem. Thus the limitation of computation.burden iswas well solved. Experiment results demonstrate that Basis Pursuit can be implemented speedy and stably, it can provide well de -noising results at the same time.