摘要
高压输电线路的导线、金具及绝缘子串等的红外图像的特点有:边缘模糊,对比度差,空间域上存在孤立的点和小块噪声。采用中值滤波方法,在保护图像边缘细节信息的前提下有效地降低了图像空间域高频噪声,用分段线性灰度变换增强了图像边缘的对比度,提高了图像的温度分辨力和空间分辨力。通过对Roberts边缘算子、Sobel边缘算子、Prewitt边缘算子、Laplace边缘算子和Laplace改进算子等常用边缘检测算子对噪声的抑制效果和边缘定位精度的对比分析,选用Roberts算子完成了图像的边缘轮廓检测,并对不连续边缘进行跟踪和连接,得到了清晰连续的图像边缘,完成了高压输电线路红外图像的边缘特征提取。以一幅耐张绝缘子串的红外图像为应用实例,验证了图像预处理及边缘检测方法的有效性。
The characteristics of infrared image for the transmission line, hardware and insulator chain in the high-voltage transmission line systems are blurry edge, blunt contrast and isolated points and nub noise in the space field. On the premise of maintaining detailed information of the image edge, by the means of median filter to effectively reduce image's noise with high frequency in the space field, and piecewise linear gray level transformation to intensify the contrast of the image edge, temperature resolving power and space resolving power of the image were improved. Then, by analyzing the restriction effect on noise and the edge location precision of several common edge detection operators like Roberts edge operator, sobel edge operator, Prewitt edge operator, Laplace edge operator and Laplace improved operator, Roberts operator was employed to achieve the image edge profile detection. After that, the clear continuous image edge was obtained by tracking and connecting the distinct edges, namely, the edge feature extraction of the high-voltage transmission line's infrared image was finally accomplished. Through the application of the infrared image of a tensioning dead-end insulator chain, the methods of the image preprocess and the edge detection in this paper is proved to be valid.
出处
《中国电力》
CSCD
北大核心
2005年第1期31-33,共3页
Electric Power
基金
国家高技术研究发展(863)计划(2002AA420110)
