A Quadric Image Segmentation for the Feature Extraction of Tire Surface Wear

As an advanced analysis tool image processing has been made a wide use in solving complex image information, such as CT, metallic erosion detect, vehicle engineering, etc. However, adopting singular image process is often constrained in its limited scope of application, and even fails in dealing. Thus we proposed quadric image segmentation based on the machine vision for extracting wear feature in this paper. The method consists of thresholding and region labeling, of which the optimization iterating of the given threshold and seed filling are used to divide the image into the multi blocks. After that the local characters could be refined with respective and independent outlines. And the corresponding cause of the formation could be interpreted judging by wear mechanism as well. Finally we compare this method with Ostu, watershed and active contour method by an instance. It shows that the method is quite a tolerant in uncertain positions, clearly segmented, quick, and fit for wear analysis