A novel point cloud edge extraction algorithm is proposed in this paper based on eight neighbor depth difference. In this algorithm, according to the point clouds of the object features, each feature point is projected vertically along the depth direction, the projection points are divided to grids, and the average of the depths corresponding to the projection points in each grid is calculated and used as the depth of the grid. Then, the depth of each grid is compared with those of its eight neighbor grids; and according to the depth difference, whether edge point exists in the grid or not is determined, and then the sorting method is used to screen out the point cloud edge point of the object in the grid. Aiming at two typical point cloud data containing nonhole and hole, the proposed eight neighbor depth difference algorithm is used to extract the edges of point clouds, which verifies the effectiveness of the algorithm.