Research on steel barrel flattened seam recognition based on machine vision

Abstract

Aiming at the problems of low efficiency and poor accuracy in manually identifying the flattened weld seam. The method of combining machine vision technology with multi-thread and multi-process is proposed, and the detection and recognition system of flattened weld seam is designed. OTSU adaptive threshold algorithm combined with Canny operator edge detection method is used for image segmentation. The probabilistic Hough transform algorithm is used to extract straight lines to identify flattened weld seam. OpenCV function library and Python are used to implement image processing algorithm, and PyQt5 is used to design human-computer interface to monitor the status of weld seam detection in real time. Meanwhile, a processing pool for parallel processing of recognition algorithm is established, and Modbus communication and image acquisition are realized by multi-thread programming. The performance of multi-core CPU is fully utilized. The experimental results show that the system meets the design requirements, has the characteristics of small amount of calculation and accurate identification, improves the processing efficiency, and solves the practical production problems.