Feature Recognition And Extraction Technology Of Debris Scattered In Traffic Accident Scene

Debris in the Traffic Accident scene contains important information about the accident. Identifying and extracting the relevant characteristics of the scattered debris in the scene of the accident can provide an important basis to solve the collision speed of the accident vehicle, also it can provide some evidence for the handing of traffic accidents.In this paper, the debris of the Traffic Accident is studied to reduce the error caused by the background in the process of calibration, which is based on the theoretical model of camera imaging and characteristics of the accident, proposed a camera calibration system combining with MATALB software. The relation model between actual value and image value of the calibration objects is established, and the red calibration objects of known dimensions is designed to obtain the photographic images containing calibration object using MATLAB software programming for image calibration object extraction. To further reduce the feature recognition process error factors, dealing with the debris image with gray transform, histogram equalization, de-noising, sharpening the image pre-processing operations; were applied Roberts, Sobel, Prewitt, LOG, Canny operators the image of fallout in two different background environments edge detection, image processing based on comparative analysis of the results. According to the comparison and analysis of image processing, this paper selected the image denoising wiener filtering, Laplacian sharpening, Canny algorithm for edge detection processing to improve the precision of image processing. In round and irregular debris as an example, the design of simulation test application proposed calibration method and image processing methods debris feature recognition, recognition result calculated and compared with the actual measurements to verify the accuracy of the proposed method.According to the similarity theory and practical conditions, car collision simulation experiment was designed which comprise car crashed barrier and two car at a certain speed. Calibration system put forward by the application of the scene of the accident analysis of image, based on the analysis of the optimal image preprocessing and edge detection algorithms write image processing program, comparing the results, the error range is 0.53% to 10.31%, there is only one debris size error is 13.59%, the reason is that the object was small, the height of the camera when shooting is relatively high, resulting in a certain depth error, and secondly because it is the method of monocular shoot, the actual background environment is not ideal, can result in large errors. By verifying through the proposed calibration system, the debris is used to identify the information, the calculation results is believable, and the operation method is simple, the method can be recommended to the traffic police department and the method expand the application range of the fallout, providing certain evidence for the scene of the accident processing.