Stone Contour Extraction And Defect Detection Based On Computer Vision

Stone is the main raw material in architectural ornament industry. Stone Contour extraction and defect detection is the chiefly step and key process. It plays a crucial role in real-time optimizing irregular stone cutting, arrange and material prepare. Besides it is important to improve production efficiency and stone use ratio, to save stone resources. Based on computer vision, this paper devises the hardware system for stone contour and defects detection. By means of analysing features of image acquisition part, grayscale, color information and texture of images, this paper proposes three algorithms for extracting stone contour, and extract stone contours accurately from stone images using the algorithms. Define stone surface detects based on its function and application. And detect regional stone defects. The main research contents and achievements are as follows:(1) Propose an overall scheme of stone detection system based on computer vision. Design the image acquisition module.(2) Extract stone contour using traditional image processing method, and analyse the problems of them.(3) Analyse the texture features in simple background images, and put forward the method that extract stone contour based on Gray Level Co-occurrence Matrix. The contours extracted from simple background images are precise using this method.(4) Define the central region in image as a sample window, and take the image information as priori knowledges of image processing. propose a method that extract stone contour based on the sample window and texture features, besides the principle about choosing key parameters of the method are analysed. For complex or simple background images, the method is accurate.(5) Via studying the color information features and comparing color spaeces, propose an algorithm based on the sample window and color information in images. In HSI color space, extract accurately the stone contour based on probability distribution similarity of H and S between the sample window and detection windows.(6) Define the stone surface defects and classify them. The regional defects are detected in this paper, such as speckle, greasy dirt, stone chalcanthite and so on. Firstly, detect the defects accurately in uniform color stone images based on histogram, but the method is difficult to detect the ones in complex stone images; Secondly, detect the defects precisely in stone images based on the method combined watershed and region merger, whose color and texture are complex.