Ground　laser　LIDAR　can　obtain　complete　and　dense　building　facade　point　cloud　data,　Based　on　such　data,　this　paper　presents　a　complete　regularization　method　of　building　contour.　Different　facades　can　be　regularized,　including　facade　walls,　doors　and　Windows,　etc.　First,　2D　α-shape　algorithm　is　used　to　roughly　extract　contour,　then　the　contour　of　the　plane　is　classified　by　Euclidean　distance　clustering,　The　RANSAC　algorithm　combined　with　the　least　square　method　is　used　to　extract　the　line　segments　for　each　small　class　of　contour　lines,　The　extracted　line　segment　is　modeled　according　to　the　main　direction　of　the　building,　and　set　a　certain　distance　threshold.　The　points　greater　than　the　distance　threshold　are　removed.　The　remaining　points　are　adjusted　in　coordinates　according　to　the　model　to　achieve　the　effect　of　regularization.　Finally,　the　regularized　point　clouds　of　each　small　segment　are　merged　to　obtain　the　regularized　contour　of　the　final　facade.　The　experiments　show　that　the　method　in　this　paper　can　achieve　good　regularization　effect　of　building　contour.　The　research　results　have　a　certain　reference　significance　for　building　3D　modeling　using　ground　laser　scanning　data.　©　2021　IEEE.