一种快速自适应边缘提取方法的研究

利用图像边缘灰度突变的特性。提出了一种梯度算子结合抽样的快速自适应边缘检测方法，在对整幅图抽样得到的低分辨率图上搜索出边缘点，然后以此为指导在整幅图上精确定位待检测的目标轮廓；为了提高边缘检测的自动化程度，将整幅图分成若干幅子图像．对每幅子图像各自统计其梯度，根据梯度信息自适应生成阈值．蛮验证明该方法不仅能快速准确地提取出单像素目标边缘而且还能较好地抑制细小边缘的于扰，在实际应用中获得了良好的效果。

Study of A Rapid Adaptive Edge- detection Method

Considering the sudden change of the gray scale of the image edge, a rapid adaptive edge-detection technique is presented which uses gradient operator and sampling. The method detects image edge in the low-resolution image which is generated by sampling in the whole image, then detects the accurate object profile in the whole image, guided by the information of the low-resolution image. In order to improve the automatization of the edge-detection, this method divides the whole image into several sub-images and counts their gradient, then works out gradient threshold adaptively according to the information of gradient, The experiment proves that this method can not only detect the single pixel profile rapidly and accurately but also avoid detecting exiguous edges, and it has good result in the application.