Abstract: To realize automatic inspection of Pandrol fastener, a Pandrol fastener spring state detection algorithm that integrates spring texture and shape features is proposed. The algorithm first divides the detection area according to the experience of manual inspection. Then the texture and contour of the fastened spring are enhanced. Then the texture features and shape features of the spring are extracted separately, and the complex feature vector is composed and put into the SVM classifier to establish the training model. Finally, the SVM classifier is used to identify the status of the track fasteners. The experimental results show that the missed detection rate is 7.1%, the false detection rate is 2.9%, and the correct rate is 92.9%, which meets the detection requirements of track fasteners.