匹配追踪算法是一种从一个极度冗余的词典中选择出某些基向量来叠加出一个特定的信号的算法 .这种算法已经成功地用于视频压缩和其它领域中 .但由于匹配追踪算法本质上是一种贪心算法 ,它的主要缺陷是无法分辨出信号中存在的双峰结构 .本文提出了一种改进算法 ,称为竞争匹配追踪算法 .这种算法能够克服上述匹配追踪算法的缺陷 ,并且在同等情况下能达到更优的性能 . Matching pursuit algorithm is an algorithm that selects certain basis vectors from an extremely redundant dictionary to superimpose a specific signal, which has been successfully used in video compression and other fields.But due to the matching pursuit algorithm Is essentially a kind of greedy algorithm whose main defect is that it can not distinguish the bimodal structure existing in the signal.This paper presents an improved algorithm called competition matching pursuit algorithm which can overcome the shortcomings of the above matching pursuit algorithm , And in the same situation can achieve better performance.