Aiming at the problems of low efficiency and high packaging error rate of USB component for manual packaging connectors, an intelligent sorting system integrating identification, selection and sorting was developed. An improved sorting algorithm was proposed,in which edge curve equivalent method was used to determine the workpiece shape, multiobjective centroid image region segmentation method was used to determine the center of each component, and Harris corner detection algorithm was used to obtain rotation angle. A binocular vision intelligent sorting and packaging mode combining sorting identification camera and rotating correction correction camera was constructed. The results show that the maximum rotation angle error is 0.75°, the maximum centroid detection coordinate deviation is 0.68 mm, and the final test success rate is more than 99.5%. The constructed intelligent sorting system can meet the design requirements of matching error and grasping misjudgment rate of the template in different positions of USB components. The research results provide reference for the research of intelligent equipment.