Machine vision recognition system for aerospace machined parts based on edge detection

Feng Chen,Xiang Shi,Haoyu Zhang,Aoyang Cai,Kechen Song,F. Yan,M. Li,X. Hou,Y. Long
DOI: https://doi.org/10.1051/e3sconf/202452201023
2024-05-07
E3S Web of Conferences
Abstract:Aerospace T-shaped machined parts are varied and have small structural differences. Manual identification has the problems of low efficiency and low accuracy. In order to realize efficient and accurate classification of aerospace machining parts, we built an image acquisition platform. To improve the edge detail extraction capability, we improved the edge detection algorithm based on deep learning. Furthermore, we employed the VisionTrain software to train recognition classification models for both large classes and subclasses. We then established a cross-granularity image classification process using VisionMaster software. Experimental results show that the improved edge detection algorithm in this paper is better than the existing common algorithm. The system achieves the goal of quickly and accurately recognizing all 60 machined parts.
What problem does this paper attempt to address?