The　precision　of　the　gyroscope　data　output　by　the　MEMS　device　of　the　ordinary　mobile　phone　is　experimentally　known　that　the　error　caused　by　the　zero　drift　in　the　stationary　state　seriously　affects　the　measurement　of　the　attitude　angle.　In　this　paper,　the　dynamic　performance　of　Kalman　filter　(KF)　is　analyzed.　A　low-cost　mobile　phone　gyroscope　denoising　method　based　on　composite　Kalman　is　proposed.　By　integrating　the　raw　data　of　gyroscope　and　accelerometer　with　large　error,　Filter　noise　reduction　processing　to　reduce　offset　drift　and　noise　and　improve　sensor　accuracy.　The　test　results　show　that　when　the　carrier　is　in　motion,　the　method　can　effectively　reduce　the　drift　error　of　the　gyroscope　and　improve　the　accuracy　of　the　attitude　angle　measurement.　After　filtering　under　dynamic　conditions,　the　attitude　angle　data　is　stable,　the　error　size　is　controlled　at　about　4%,　the　error　is　within　the　acceptable　range,　and　the　comprehensive　noise　reduction　effect　is　about　77.36%,　which　proves　that　the　algorithm　can　effectively　improve　the　measurement　accuracy　of　low-cost　MEMS　gyroscopes.　©　2019　Computer　Society　of　the　Republic　of　China.　All　rights　reserved.