Automated visual inspection of multipattern train components using gradient information and feature fusion under the illumination-variant condition

The condition of train components needs to be inspected periodically to ensure the quality and safe operation of trains. The brake beam bolt is an important locking part in the main structure of the support and connection components. Any fault in the break beam bolt can lead to serious accidents. With the development of computer vision techniques, automated visual inspection methods have started replacing routine manual checking. This paper presents an automated and hierarchical inspection method for the detection of the condition of the brake beam bolt. Gradient information and basic feature descriptors are fused into a synthesized feature expression for eliminating the influence of the illumination-variant condition. The fused feature is also employed in the identification of the object condition. The results of the experiment show that the feature fusion expression form can effectively solve the problem caused by the illumination-variant scene and condition. The correct rates of localization of joint part and target fault detection have shown high accuracy. It has been concluded that the overall performance of the visual inspection system can meet the practical object safety assurance application.