Visual Recognition of Traffic Accident Scene’s Image Research

The optimization of the visual image of traffic accident scene can effectively improve the efficiency of traffic accident treatment. To identify the visual image, we need to obtain the morphological characteristics of the scene image, and establish the sector feature vector of the scene of the traffic accident image to optimize the image recognition. The traditional method extracts a reasonable threshold by extracting the neighborhood image features of the key points of the image, but ignores the establishment of the image sector region feature vectors, resulting in low recognition accuracy. Intelligent vision based visual image recognition method for traffic accident scene. The scene of the accident to extract the background image using mean background modeling theory, the background image is normalized, and the accident scene contour image into two value image, the traffic accident scene image filtering, to obtain morphological characteristics of scene images of different sizes, the establishment of traffic accident scene image regional feature vector, combined with local Discrimination Mapping Theory dimensionality reduction of image feature vectors of traffic accident scene, complete the traffic accident scene visual image recognition. Experimental results show that the proposed method can effectively improve the visual image recognition rate of traffic accident scene, and the recognition accuracy is high.