The purpose of this work is to develop an algorithm for recognizing shrinkage defects on metal alloys. The steps for image processing are described. When analyzing the algorithm, a description is presented for each step, followed by a software implementation. The final step, after using the methods of the algorithm, is to count the contours of the defects. To demonstrate the functionality of the software, random photos of the metal body in the section were taken, where defects in the form of gas pores can be directly observed. The software, with the correct selection of input data, processes the image with high accuracy, minimizing errors in calculating defects. Using the proposed algorithms reduces the diagnostic time. The article presents a comparative description of the manual and automated methods, showing the effectiveness of the second method in comparison with the first. To write the software, the Python 3.7 programming language was used, as well as the OpenCV computer vision algorithms library.

Keywords: defect detection, computer vision algorithms, OpenCV, metal defects, shrinkage defects