Compressed　sensing　theory　is　a　new　research　direction　in　the　field　of　modern　signal　processing,　and　direction　of　arrival　(DOA)　estimation　is　a　research　hotspot　of　this　theory.　In　practical　engineering　applications,　the　traditional　DOA　estimation　algorithm　usually　needs　to　collect　a　large　number　of　independent　snapshot　data,　and　the　estimation　accuracy　is　low　and　the　calculation　amount　is　large　when　the　observation　data　and　dictionary　matrix　contain　noise　at　the　same　time.　In　order　to　solve　the　above　problems　effectively,　this　paper　presents　an　improved　algorithm　based　on　Regularized-FOCaI　Underdetermined　System　Solver　to　Multiple　Measurement　Vectors　(RM-FOCUSS).　The　algorithm　obtains　higher　estimation　accuracy　and　angle　resolution　by　iterating　between　sparse　solution　and　dictionary　error.　The　simulation　results　verify　the　reliability　of　the　algorithm.　©　2019　IEEE.