Adaptive image quality correction for workpiece contour extraction and positioning in unstructured scenes

In automated operation scenarios such as mixed-flow sorting and workpiece polishing, it is often necessary to obtain contour extraction and accurate positioning of workpieces. However, there is a general problem that the lighting environment is changeable, which makes it difficult to form a structured scene. In addition, the parameters of the camera cannot be adjusted in real time, which leads to the poor imaging quality of the workpiece, thus affecting the positioning accuracy of the workpiece and the results of contour extraction. Therefore, it is necessary to research workpiece positioning and contour extraction methods in unstructured scenes, including the following three points: 1) internal and external parameter calibration of the camera; 2) evaluating and improving workpiece imaging quality; 3) workpiece contour extraction and positioning. Finally, experiments were carried out in different illumination scenes, shape sizes and surface materials, obtaining positioning errors within 1 mm, and verifying the effectiveness and practicality of the proposed method.