Abstract: In view of the problems of ‘blind spots’ in complicated goaf detecting by using laser scanning and point cloud density distribution inhomogeneity, this article introduced multi-point laser scan and point cloud merging and compression. Multi-point scan in complicated goaf avoided ‘blind spots’ and densified sparse point cloud regions. The merging algorithm of point cloud data was put forward based on a common coordinate system and the least-squares principle to solve the target transformation matrix. After the distribution rule of point cloud concentration areas was analyzed, the scattered point cloud compression algorithm was proposed, in which the point cloud was divided into portions along the y direction firstly, then intralayer data were divided by the extreme values of x and z, and each point was sorted on the x value and screened on step k. Error analysis of an instance of large versed goaf shows that the merging algorithm based on the least-squares principle will achieve high precision with an error range of about 0.1 mm. The compression algorithm can achieve a compression proportion of 15% to 25% and ensure the integrity of 3D boundary information at the same time.