反馈型颜色分割在边缘检测中的应用

Application of Feedback Color Segmentation in Edge Detection

针对传统的边缘检测算子容易产生边缘丢失和畸变的缺点,为准确获取彩色图像的边缘,本文提出了将传统方法提取的边缘作为反馈对原始彩色图像进行初步颜色分割,根据区域内颜色信息的分布统计,进行分裂合并及均衡化处理,对颜色区域边界进行骨架细化,从而提取边缘。实验结果表明:与传统方法比较,该方法可以更有效地检测出边缘信息。

In edge detection, lose and distortion of edges will happen with traditional operators which are based on analysis of image grayscales. In order to accurately detect edges of color images, a new method is based on the feedback system. Initial color segmentation will be finished using the result of edge detection with traditional method as a feedback on color image. According to the distribution statistic of color information, color regions processed with equalization are separated independent to extract edges after skeleton thinning. It is proved that this new method can effectively detect edges of color images compared with traditional methods.