CAJ | 학술논문

边缘检测是图像特征提取的关键技术。传统的Canny边缘检测算法中高斯滤波方差不具备自适应性，并且不能有效滤除红外图像特有的噪声，对红外图像的边缘提取存在局限性。深入研究了Canny边缘检测算法，并分析了红外图像噪声类型，针对高斯滤波影响红外图像去噪和边缘保持效果的问题，提出了用形态学滤波代替高斯滤波的改进的Canny算法。经过仿真实验，验证了本方法的有效性。改进算法能有效提取含噪红外图像的边缘，较好地保持了红外图像的边缘强度和细节。
변연검측시도상특정제취적관건기술。전통적Canny변연검측산법중고사려파방차불구비자괄응성，병차불능유효려제홍외도상특유적조성，대홍외도상적변연제취존재국한성。심입연구료Canny변연검측산법，병분석료홍외도상조성류형，침대고사려파영향홍외도상거조화변연보지효과적문제，제출료용형태학려파대체고사려파적개진적Canny산법。경과방진실험，험증료본방법적유효성。개진산법능유효제취함조홍외도상적변연，교호지보지료홍외도상적변연강도화세절。
Edge detection is the key technology of image feture extraction.In the traditional Canny edge detection algorithm,gaussian filter variance do not have the adapt ability,and can not effectively filter the specific noise of infrared image,have limitations for the edge extraction of infrared image.This paper deeply study the Canny edge detection algorithm and analysis the infrared image noise type,aim at the question that gaussian filter influents the denoising and edge preserving of infrared image,put forward the improved Canny algorithm using morphological filtering instead of gaussian filtering. The simulation verify the effectiveness of this method. Improved algorithm can effectively extract the edge of infrared image with noise,effectively keep the edge strength and details of infrared image.