摘要
边缘检测对光照强度的反应很敏感,同一种算法很难对不同光照情况下的图像进行有效的边缘检测。通过对不同光照情况下的图像特征分析,提出了不同光照情况下的图像边缘检测算法:正常光照条件下采用Sobel算子进行边缘检测;弱光照条件时先用MSR(multi-scaleretine)进行图像细节增强处理,再利用类别方差自动门限法得到图像边缘;强光照条件下通过灰度扩展,将所感兴趣的灰度范围放大得到图像的边缘。实例分析表明,该方法能够从灰度图像中获取较为理想的图像边缘,并具有一定的抗噪性能。
Image edge detection is sensitive to different illumination conditions. The characters of different illumination conditions are studied and the corresponding algorithms of gray image edge detection are proposed. In normal condition, the Sobel operator can be used to get the image edge. When the illumination is faint, the image is first enhanced by the MSR, and then the variance threshold algorithm is employed. While the light is strong, the image edge can be got by gray expand. At last, the instance image is detected. It is shown that the algorithms are very effective in different illumination conditions, and can resist noise.
出处
《北京机械工业学院学报》
2006年第3期25-27,38,共4页
Journal of Beijing Institute of Machinery
基金
北京市教委科技发展计划面上项目(KM200611232002)
