基于改进区域生长法的电力设备红外图像分割

为实现电力设备红外图像中目标设备图像的分割,对传统区域生长法需要人工选择初始种子点,以及易产生过分割与欠分割的不足提出了改进,将红外图像中邻域均值最大的像素点作为种子点,实现种子点的自动选取。提出了使用Sobel算子计算梯度幅值作为附加限定条件的生长准则。改进的区域生长法的分割效果比传统的区域生长法效果更好。

Mentation based on modified region growing algorithm

In order to segment the target power equipment in the infrared image, according to the drawbacks of traditional region growing algorithm that the initial seed points have to be selected artificially and it is easy to cause under-segmentation and over-segmentation, the region growing algorithm is modified.The pixel point with maximal neighborhood mean as the initial seed point is selected to realize selecting the initial seed point automatically.The Sobel operator to calculate the gradient magnitude is used as the rule of growth of the additional limiter.The effect of the modified region growing algorithm is better than the traditional region growing algorithm''s.