摘要
提出一种基于Bezier曲线的红外图像边缘检测的循环移位算法。为抑制红外图像中的噪声的影响,利用Bezier曲线法平滑图像灰度直方图中的噪声,得到Bezier直方图。利用提出的“循环移位法”探测Bezier直方图曲率曲线的极大和极小值,确定分割阈值和量化灰度值,实现对原始红外图像的分割,对分割后的图像进行边缘检测。实验结果表明,新算法简单有效,在准确检测红外图像边缘信息的同时又能抑制图像中噪声的影响,定位精度高,而且能够得到单像素边缘,在性能上优于传统的几种边缘检测算子。
A new method for edge detection of IR image based on Bezier curve is proposed. In order to inhibit noise influence on IR image, the noise in image gray scale histogram is smoothed by Bezier curve to obtain Bezier histogram. Based on the proposed circle-shift method, the maximum and minimum of Bezier histogram curvature curves is detected. The segmentation threshold and the quantitative gray level can be obtained and the original IR image can be segmented. Edge is detected for the segmented images. Experimental results show that the new algorithm is simple and effective to detect the whole contour and detail information while suppressing noises. Location accuracy is high and single pixel width edge can be detected.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2005年第5期27-30,共4页
Opto-Electronic Engineering
