In order to improve the recognition efficiency of traffic road centerline recognition method, a traffic road centerline recognition method based on speech recognition technology and image feature extraction is designed. Firstly, the remote sensing image is preprocessed, and then the road knowledge base of traffic road image is established, which mainly includes four parts: road feature analysis, building road extraction knowledge base, high-resolution remote sensing image road rule set and traffic road image classification. On this basis, the image is classified, and finally the road midline feature points are extracted by proportion space theory to realize the final intersection Central line identification of access road. The experimental results show that the recognition time of the proposed method is less than that of the traditional method, and the recognition accuracy is high, which has a certain practical significance.