Abstract
An automated tire mark identification system includes functions of the mark localization, segmentation, feature extraction, tire-tread matching and identification. The tire mark localization is the first and most important step for the system development. The study proposed a new algorithm called rotation-projection method to solve the problem of the seed selection by using region growing algorithm for skid mark localization and expand the processing scope of the types of tire marks. This new algorithm has six steps including gradient direction angle detection, image rotation, image projection, calculating the coefficient of the variant, and detecting the tire mark area. This method combined with the gradient operator can effectively detect the area of the new or old tire marks. Its accuracy is similar to the original region growing algorithm in new tire mark localization. Nevertheless, the algorithm also can localize the old tire mark while the region growing method can not cope with.
