Implementation and Design of Metro Process Quality Inspection System based on Image Processing Technology

Abstract

In recent years, China’s subway industry has entered a phase of rapid development, and at the same time, the commissioning of large-scale new lines has brought huge challenges to production and operation. In the quality inspection system of subway process, the traditional way is to use human as the carrier of quality inspection, but with the input of large-scale production lines, the manual inspection method is far from meeting the needs of industrial production. Based on the above background, this paper applies image processing technology to the quality inspection process of subway process, and completes the design and implementation of the quality inspection system of subway process based on image processing. In this paper, a lightweight feature-enhanced convolutional neural network model is proposed under the premise of ensuring efficiency. The model deepens the structure of the backbone feature extraction network, adds another prediction scale layer, realizes three-scale prediction, and enhances the semantic information of the shallow feature map, thus improving the detection accuracy of small-scale defects. At the same time, the algorithm has room for optimization and improvement in “identification of anti-loosening wire” and “accurate positioning of through-hole gaskets and washers without missing”.