Research On Graphical Analysis Method For Tyre Wear

Not only being the important part to constitute the running gear of vehicle, tyre is but the only body through which vehicle is in contact with the ground. Whether good or bad tyre's performance is has direct effects on its durability and stability, even more vehicle's safe, saving, controllability, riding comfort, and noise etc. It is thus very meaningful to make a study on tyre wear. During the driving, tyre wear's strength, region and shape must be reflected in the tread texture, based on that, a graphical analysis method for wear state evaluation on running tyre is presented in this dissertation, and a study for wear analysis on single image and image series are deployed.(1) By gradient operator, LoG operator, canny operator, texture's edge is detected, while the texture's region being detected by watershed method, active contour method, and texture's gray scale threshold detected by iteration threshold method and Otsu method. After that, the corresponding different characters are respectively obtained about dealing such parameters as edge for referential position, region for identifying patch, and gray scale threshold for contrast background with object. Due to that, it is denoted that an algorithm for detecting wear should be designed according to the features of tread texture.(2) Aiming at detecting and quantifying wear texture in comparatively high resolution, we present region labeling method based on morphology. Through mask designing, region detecting, region tracing and region quantifying, regions with clearer boundary are extracted, at the same time their corresponding values of region features, e.g. area, perimeter, and Centroid, etc are calculated out. To quicken the speed on region's detection and quantification, Otsu is introduced. And the method that extracting and quantifying texture features by region labeling incorporated with Otsu is presented. Based on region's area and position, a criterion on judging whether the detected region is a wear one is presented. And abnormal wear region and wear state is therefore analyzed out.(3) In early time of the wearing, tread texture is with an ambiguous character. Due to that we present an approach based on wavelet transform to analyze the tread wear texture. According to the symmetry of pattern composition, wavelet function based on Quadratic B-spline is chosen to extract the edges as the mutated points in gray scale. To coastdown the conglutination and breakage, through analysis on texture's shape features as region's interior angle, corners, circularity, etc, similarity calculation is presented to make a local match between the dealt results from by wavelet and by Otsu, so that the contour extraction could be modified better. To quicken the searching procedure in the wavelet transform analysis, a weighted involute searching method is presented. And such parameters as occupying ratio of gray scale in wear region, regional distribution in gray scale, and so on, are proposed as indices for analyzing tyre wear.(4) For describing the tyre's states and providing data for wear analysis, after comparing the results of calculation amount and smoothness of surface approaching when constructing the same surface by Coons, Ferguson and Bezier, we present a modeling method based on Bezier with metas embedded into skeleton. The whole model consists of basic skeleton to control contour and patches to be filled with, while the former is formed by Bezier spline, and the latter being tensed and dilated by Bezier surfaces. Simulation system is designed by VB & OpenGL. In this system tyre modeling, tread texture mapping, texture capturing, feature extraction, and also morphological operation on pixels are all realized. And this system is the base of model aided wear analysis.(5) For analyzing on tyre wear image and image series, in turn for mastering the progress of wearing variation, we present model aided analysis on tyre wear. And the model aided analysis system including object modeling, wear analysis, data pool and data administration is constructed. Image series are generated by the existed wear texture images, and following that, the wear's varying course is discussed by the theory on nonlinear time series. By auto correlation function the delaying timeτis calculated, and the embedded dimension m is calculated by FNN-D. Byτand m, the phase space is reconstructed, from which the maximum Lyapunov exponent L~* is calculated by small data sets. The difference in region's variation between image series is found by image series match. Based on those the varying trend of wear is discussed. The result shows that the tyre wear texture image series has chaotic tendency, with being sensitive to wearing direction.