Detection Method Of Timber Defects Based On Computer Vision

Throughout the history of the use of wood, wood because of their natural texture and color to be used in a large number of industry. Applications range of wood from wood carving to large wooden building. In recent years, some special textures species rapid rise in prices is due to the people’s favorite. With the development of the wood processing industry, quality inspection of wood became the first premise of the decision of the timber value. Due to limitations of the human physique and increase labor costs, Computer Vision Detection as a new means of detection appear in the automation process.For this article, wood-based detection is based on computer vision. After obtaining wood image, we research algorithm by studying the characteristics of wood defect image. Through the study we propose two types of images based on different characteristics of defect segmentation.Two research using threshold segmentation algorithm in a grayscale image:One is the defect image threshold segmentation algorithm based on defect-free standard templates timber. Artificial selection defect-free standard templates, establish a standard gray value. By comparing the test sheet images and templates of all the pixels, segmentation threshold is the average of the sum of all the points difference and the value; The second kind is based on a single pixel defect threshold image segmentation algorithm timber. No manual selection defect-free standard templates process. Use the difference between two pixels is determined the defect boundaries and a normal region the texture region. So as to achieve the extraction of wood defect image segmentation purposesIn the color image as the background of the premise, the images are converted into image-based HSI color space and LAB color space, and on both images using the K-means clustering method to defect segmentation. Finally, the two images chang into based on RGB color space. The combined use of regional consolidation algorithms determine region. The combined image using mathematical morphology operation noise filtering, to obtain the final image segmentation.