基于CT图像板材节子边缘检测的研究

摘要

Edge extraction of panel CT image is key step to achieve panel three-dimensional reconstruction. Aiming at problems of low contrast , low dynamic range characteristics and difficulties in panel defects edge extraction, the edge of panel image treated by binary was extracted by using mathematical morphology, also, a comparison was made between our new method and the traditional Sobel, Prewitt, LoG and Canny operation. The results showed that, our method can detect the edge information of panel knot and background , extract the edge clearly and completely, and with high accuracy,thus,the method can establish basis for subsequent knot feature points extraction and three-dimensional reconstruction.%板材CT图像的边缘提取是实现板材三维重建的关键步骤.针对板材CT图像低对比度、低动态范围、板材缺陷边缘提取困难等问题,采用数学形态学方法对板材二值化处理后的图像进行边缘提取,并与传统的Sobel、Prewitt、LoG和Canny算子对原始板材CT图像边缘提取效果进行对比.结果表明,这种方法能较好地检测出板材节子的边缘信息,提取的边缘清晰完整,准确性较高,能为后续的板材CT图像三维重构奠定基础.