Abstract: A novel sparsity adaptive orthogonal complementary matching pursuit algorithm is proposed when the sparsity is unknown. Signal is reconstructed by complementary orthogonal matching pursuit through initializing the sparsity, and a signal support can be determined. If the condition is not be reached, the sparsity is increased with specified steps, and the algorithm is re-used to reconstruct signal. The algorithm terminates when the condition is reached and the support is founded. Experimental results show that signal can be reconstructed accurately and effectively. Meanwhile, the proposed algorithm exhibits its superiority over other algorithms with the same compressed ratio.

Key words: sparsity, orthogonal complementary, specified steps, support

摘要： 针对实际应用中信号稀疏度未知的缺点，提出了一种稀疏度自适应的正交互补匹配追踪算法。算法先初始化稀疏度，再通过互补正交匹配追踪重构信号，找到一个支撑集；若支撑集不满足条件，则按指定步长增加稀疏度，再次运用算法进行重构，直到支撑集满足条件，得到最优支撑集。实验结果表明，该算法可以准确有效地重构信号，并且在相同压缩比下，其重构质量（PSNR）优于其他几种算法。

关键词: 稀疏度, 正交互补, 指定步长, 支撑集