Study On Tire Marks Image Processing And Recognition Algorithm

With the development of social economy and sharp increase of the number of car ownership, people have more and more persecution coming from traffic jams, traffic accidents-happened frequently, air pollution, serious damage of roads etc., while people enjoy the convenience and profit brought by the mobility. At the same time, the number of the crimes by motor vehicles and traffic accident increases sharply. In these crime scenes, the tire marks on the road surfaces, skins, clothes, textile fabrics and so on, can often be found, confirmed and collected. The tire marks in crime scene are the important evidence of traces. It is also the key to find the vehicle in the case or suspicion vehicle. Then the tire trace evidences are very important in crime or accident investigation.Fuzzy recognition of the tire marks are focused based on mathematical morphology. It is significant in the field of image processing and analysis on tire traces. The recognition and segmentation on the edge of the tire tread pattern and individual features in image of tire mark is the most important and most difficult in the research on the tread pattern block structure and tread pattern change of form about image of tire marks. It is badly in need of the quantitative analysis of the edge detection and features division in the image of tire marks. One of basic means in the identification of the tire marks is the features division of tire mark images. It is also the basic means from qualitative analysis to quantitative one in this field. Based on picture theory and computer machine vision technology, the system of digital picture fuzzy recognition of tire marks can be uses to recognize the morphology of the tire tread pattern and confirm the form of the tire tread pattern, facilitating to distinguish the kind of vehicles and reduce the scope of the vehicles in the cases. It is significant in recognizing the kind of vehicles, detecting and collecting the image edge of tire marks, comparing and measuring the trace sample and stamp preparation sample, confirming the suspicion vehicles, etc. It explores the issues such as detection of the edge, collecting the features, features matching, fuzzy recognition, and so on. Some innovations are given in analyzing the algorithm. This dissertation takes into the consideration of multidisciplinary fields and up-to-date technology, and the main contents as follows:1.With the help of trace recognition theory, the first part introduces the conception, features and basic theory on trace recognition in details; describes the features of tire marks by introducing the structures, types, standards of tires; analyzing the theory of tire trace shaping; and giving the definition and form features of tire tread patterns in accidents scenes.2. The preprocessing of tire mark images solves the shortage of collecting means of the images of tire marks in accidents scenes. By tests, gray level images changes in adjusted value being 3, gives the best effect. With the help of the image smoothing, filter, sharpen programs in the MATLAB, it keeps the original information of the images of tire mark, enhancing the definition of the images of tire mark, and preparing for the edge detection and collecting the features for the further part in the dissertation.3. After comparing with the traditional algorithms, this paper puts forward the method of morphology optimizing edge detection based on the canny algorithms in order to analyze three classical images of tire mark. To color images of tire mark in accidents scenes, this dissertation takes color space model to convert, and realize detection of tire mark image edges by contrasting color difference.4. Taking the advantages of multi-resolution analysis and discrete wavelet, this dissertation decomposes and reconstitutes the wavelet of the images of tire marks, then gets the features of multi-resolution wavelet of the images of tire marks in different scales. So this dissertation puts forward a watershed mesh segmentation algorithm based on label control in morphological reconstruction technique. Processing by this algorithm, it improves the working speed and reducing the complexity. This dissertation also proposes the way to improve watershed algorithm. By label control based on morphological reconstruction technique, it signs foreground and background, and reduces the over-divided questions.5. With the algorithm of images recognition in this paper, in the last part takes the simulation experiment to kinds of the images of tire marks, and matches comparison the image features of tire marks. And at the same time, it exploits Visual FoxPro programming to inquire out certain tire tread pattern. The database of tire tread pattern collects more than 500 pictures of tire tread patterns and almost 10,000 data. With the data software, the time of comparing the data of tire tread pattern and confirming the scope of vehicles in cases will be reduced clearly.The tire marks have well stability and specificability. As fingerprints of the vehicles, the tire marks have unique advantages to get the vehicle types and the vehicle information. The results in forensic science are very important.With the development of science and technology and the in-depth research, it will be significant in the field of disposing and fixating the liability of the road traffic accidents. On the other hand, the research and exploitation on trace evidences is a classic example of combining the image processing technology and crime scene investigation, and it acquired well effect on realizing technology and algorithm. The conclusion of this dissertation will bring the gains of both theory and practice in the image processing technology in crime investigation.