Abstract: Aiming at the problems of inefficient manual sampling of irregular pressed parts and easy visual fatigue, a multi-feature and multiple detection method for pressed parts based on HALCON vision software is proposed. According to multiple features of the contour of the pressed part, this method uses image gray-scale and morphological processing, uses gray-scale stretching algorithm, iterative threshold algorithm, etc. to extract and segment the contour, and judge the attributes of the segmented contour feature, and take the minimum The two-fold fitting method can accurately detect the features of the pressed parts. Collect 100 pressed products and use the above-mentioned method to detect, and the detection accuracy rate reaches 97%. Experiments show that this pressing part detection method is suitable for pressing parts with complex and multiple features. It has a high tolerance to the shape of the pressed parts and a fast detection rate, which meets the high-efficiency inspection requirements for such precision products in the industry.