Abstract: Multi-view laser point cloud registration is the basis of three dimension reconstruction, and the extraction of overlapping regions in multi-viewpoint laser point clouds is of great values to improve the efficiency of laser point cloud registration. A method of overlapping regions extraction based on region segmentation was presented, the spectral clustering was used to segment the point clouds of each viewpoint according to the geometric structure, and then a multi-dimensional shape descriptor was created for each region. The Euclidean distances were calculated for each extracted descriptor, the area with nearest Euclidean distance between descriptors was the overlapping area between point clouds. Experiments show that the algorithm is stable to the laser point clouds noise and the initial position, and the algorithm could still complete the extraction of overlapping regions in the case of large differences between point clouds. With the simulated multi-viewpoint point clouds, the overlap ratio increased by an average of 14.3%. And with the actual multi-viewpoint point clouds, the overlap ratio increased by an average of 13.3%.