In order to improve the effectiveness of defect recognition and enhance the quality of mechanical parts, a laser 3D point cloud based surface defect recognition method for mechanical parts is proposed. Using HandySCAN 3D laser scanner to collect three-dimensional point cloud data on the surface of mechanical parts, and removing external interference through statistical filtering; Using the Sampling Consistency Initial Registration (SAC-IA) method and the improved ICP method to achieve coarse and fine registration of 3D point cloud data on the surface of mechanical parts, aligning point cloud data from different perspectives and lighting conditions, and providing complete and consistent 3D point cloud data on the surface of mechanical parts for subsequent defect recognition; By training point cloud data using an unsupervised defect detection method based on GAN, the features of surface defects on mechanical parts are automatically learned, achieving accurate recognition of surface defects on mechanical parts. The experimental results show that the proposed method has high recognition accuracy and adaptability.