| Tree-ring is composed of homocentric circles, and every lap of circle represents the tree's growth information in a year. Counting the ring number by human is a time-consuming job. Recent years, with the development of Image Processing technology, it is possible to count the ring number of trees automatically and faster.In this paper, we used digital Image Processing technology to analyze tree-ring photos to extract the information of tree-rings and get trees age faster. We first use Matlab software to read tree-ring image and convert the image to a gray-scale image. Then, we process the gray-scale image with image enhancement to eliminate the image noise. Finally, the ring lines can be extracted from the image. Sobel filter, Prewitt filter, Roberts filter, LOG filter, Canny filters and some other filters have be used to extract the ring lines. The detailed treatment processes of these filters are introduced, and the extract effects of these filters are compared. The comparison results show that Canny filter is the best for ring line extraction, which can get the clear and integrity ring lines with less image noise facilitating the extraction of tree-ring information. Therefore, in the paper, canny filter is choosed for ring line extraction. However, tree-rings often have defects, such as capsules, section scar and cleft, which makes the processed image still has noise in some regions. This kind of noise is inevitable, and corresponding methods must be adopted when extract ring lines. In the paper, we first use the region-selection method to divide the parts whose ring lines are clear and without defects from the whole image; then, we use binaryzation image method to make the ring lines of divided parts more obviously; finally, the scanning-line approach is used to process the divided image to get satisfied results. We have experimented with more than 20 ring images. Experiment results show that our method can get ring number effectively and the detection correct rate is more than 98%. |
PDF Full Text Request