当图像中的某些区域具有比其它部分更高的重要性时, 基于感兴趣区域(ROI)的边缘检测功能就尤为重要。但是当前大部分算法均针对图像整体检测, 这在一定程度上会影响对ROI区域的检测效果。针对该问题, 提出一种基于ROI的边缘检测算法。新算法首先利用修正后的图像直方图特征选取ROI区域的分割阈值, 然后根据分割阈值从背景中分离出ROI区域, 最后选择最优边缘检测算子, 完成基于ROI的边缘检测。实验结果表明：新算法能够更好的支持对ROI区域的边缘抽取。

The functionality of region-of interest(ROI) image edge detection is very important in applications where certain parts of the image are of higher importance than others. However, at present, most algorithms are designed for the whole image detection, which will affect the ROI edge detection performance. In this paper, an image edge detection algorithm based on ROI extraction is presented. The new algorithm first chooses the ROI partitioning threshold using the proposed histogram modification. Then, it segments the ROI from the whole image according to the new partitioning threshold. Finally, an optimal edge detection operator is selected to complete the ROI edge detection. Experimental results show that the new algorithm can better support the ROI edge extraction than those whole-image detection operators.