关键词: 螺栓检测, 特征提取, 图像梯度, 支持向量机, 动车

Abstract: Bolt, which lies at the bottom of China railway high-speed, is a key component of train braking system, and plays a significant role in the safe braking and running of train. The traditional locomotive maintenance mode, which is carried out by trained workers, is no longer adapted to the train maintenance needs of high-efficiency and high quality. With the advancement of computer and electronic technology, the online inspection based on machine vision has played more and more important role in industrial field. Under the outer complex environment, the automatic inspection for bolt of CRH using image processing and analysis is full of challenges. In this paper, a method based on feature extraction and machine learning is presented, which fulfills the fast localization of bolt. Experiment result has demonstrated that the proposed algorithm is very robust to outer complex environment, especially to the changes of illumination.

Key words: bolt detection, feature extraction, image gradient, support vector machine, China Railway High-speed