Aiming at the difficulty of scene segmentation in the process of robotic random bin picking, a point cloud segmentation method based on the improved Euclidean clustering is proposed. The pass-through filter and the iterative radius filtering are used for the pretreatment to obtain the point cloud of scattered workpieces after removing the interference points. The edge points in the point cloud are removed by the edge detection based on normal angle, and the inter-collision workpieces are separated in space. The improved radius adaptive Euclidean clustering is adopted for the point cloud segmentation to obtain the point cloud subsets of many workpieces. The removed edge points will be put into the point cloud subsets based on the distance constraint, and thus the point cloud segmentation is completed. In addition, the offline template point cloud provides reference for the selection of segmentation parameters, which ensures the accuracy of segmentation results and improves the segmentation speed. The experimental results show that the proposed method can accurately segment the interested workpieces, and the segmentation time is about 696 ms. It is satisfied with the real-time requirement of industrial robot picking.