压缩采样匹配追踪(CompressiveSampling MP)是D. Needell继ROMP之后提出的又一个具有较大影响力的重构算法。CoSaMP也是对OMP的一种改进，每次迭代选择多个原子，除了原子的选择标准之外，它有一点不同于ROMP：ROMP每次迭代已经选择的原子会一直保留，而CoSaMP每次迭代选择的原子在下次迭代中可能会被抛弃。(CompressiveSampling MP is another influential reconstruction algorithm proposed by D. Needell after ROMP. CoSaMP is also an improvement over OMP. Each iteration selects multiple atoms. In addition to the selection criteria for atoms, it differs from ROMP in that ROMP's already selected atoms are retained for each iteration, and CoSaMP selects each iteration. The atom may be abandoned in the next iteration.)