This　paper　proposed　an　improved　method　which　can　reduce　the　misclassification　in　human　pose　estimation　based　on　random　forest　and　increase　the　accuracy,　included　adaptive　fusion　feature　extraction　and　misclassification　processing　mechanism.　Firstly,　we　improved　the　method　of　feature　extraction　to　adaptive　extract　deep　fusion　feature　with　adaptive　feature　fusion　extractive　method,　so　that,　both　distance　information　and　part　information　could　enhance　feature　expression.　Furthermore,　owing　to　inspiration　from　error　cluster　analysis　and　iteration　thought,　the　misclassification　processing　mechanism　is　proposed　to　handle　misclassification　appearance.　Finally,　we　achieved　accurate　human　pose　estimation　from　single　depth　images　by　applying　the　principal　direction　vector　based　on　the　improved　principal　direction　analysis　(PDA)　algorithm.　The　experimental　results　demonstrated　that　this　algorithm　can　efficiently　eliminate　several　misclassifications　and　improve　the　accuracy　of　the　3D　pose　estimation.　Copyright　©　2020　Acta　Automatica　Sinica.　All　rights　reserved.