Because　the　traditional　two-dimensional　direction　of　arrival　(2D　DOA)　estimation　algorithm　has　low　estimation　accuracy　and　high　calculation　complexity　under　the　condition　of　low　signal-to-noise　ratio　(SNR)　and　few　snapshots,　this　paper　studies　the　2D　DOA　estimation　algorithm　under　compressed　sensing　theory.　We　first　reduce　the　dimensionality　through　the　spatial　synthesis　angles　in　L　array,　then　construct　the　sparse　reconstruction　models　separately,　use　the　improved　orthogonal　matching　pursuit　(IOMP)　algorithm　to　reconstruct　two　sets　of　synthetic　angles,　and　finally　through　an　angle　matching　algorithm　based　on　orthogonal　matching　to　get　the　pitch　and　azimuth　of　target　source.　The　simulation　results　show　that　the　algorithm　in　this　paper　has　higher　estimation　accuracy　than　the　traditional　algorithm　under　the　conditions　of　less　snapshots　and　low　SNR.　It　does　not　need　to　know　the　number　of　signal　sources　in　advance,　and　can　also　obtain　the　correct　angle　pairing　when　the　amplitudes　between　signal　sources　are　similar,　which　promotes　the　further　fusion　of　compressed　sensing　and　DOA　estimation　under　actual　conditions.　©　2020　ACM.