Tire Prints Image Enhancement Method

With automobile user's quantity of our country increase annually, car accidents and hit-and-run cases also be growing day by day. So vehicle imprints has become one of the key focus in the handling and detection of these cases. With the development of computer science, especially the use of digital image processing technology of tire prints collected at the scene of the accident, which has become important method of responsibility of road accidents and identification of hit-and-run vehicle, as well as analysis of the accident cause. And which is an effective method of science for achieving an objective, efficient traffic processing.Tire imprints image enhancement is the pretreatment technology for the image feature extraction and recognition of tire imprints in the accident. Because tire imprint images of the scene are affected by external environment, weather, light and other factors, most of images associate with low contrast, noise, and other characteristics, resulting in the indistinctly individual features in the tread pattern and a big gap between visual look and actual needs. Which is unfavorable to further processing study. Therefore, it is necessary to enhance the tire imprint images. At present most of enhancement methods of tire imprints image is only a simple linear transformation, gray stretch, noise reduction etc., which often lead to the distortion of the image color, the loss of detail information. So the effect of these enhancement methods is not very satisfactory. Based on this, we propose more professional image enhancement methods of tire imprints, and which are proved by experiments in the text and can achieve better enhancement results, which has a strong theoretical significance and application value too the detection of accident vehicle of cases.This paper introduces the background and significance of the vehicle tire imprints, and describes research status at home and abroad,and that there are some problems with these research methods. Then it has a simple introduction to some basic knowledge of tire imprints and reviews processing techniques of tire imprints.The following are the main contents of this paper:(1) According to there is uneven distribution of light intensity in the tire imprints image, we propose multi-scale Retinex to enhance the image by adjusting the image's dynamic range, color constancy, image sharpening and so on. The paper also studies the computational model of Retinex theory deeply, and points out that there are some limitations on the tire imprints enhancement for single-scale Retinex. and proves that there are some advantages for multi-scale Retinex algorithm enhancement method by experiment.(2) According to there are a lot of noise, uneven distribution of light sources and other issues, we propose an image enhancement basing on wavelet decomposition and reconstruction. By using Mallat fast wavelet decomposition algorithm, the image of tire prints is decomposed into a number of different scales of high and low frequency components, and then which are treated with different methods, overcoming the limitations of traditional methods, which can only remove the image noise or only adjust the brightness distribution of image.