基于ORB和最小凸包的感兴趣区域检测方法研究*

随着科技的发展，如何准确检测出复杂背景情况下的感兴趣区域(ROI)和提高检测方法的实时性已经成为图像处理领域亟待解决的问题。针对此问题，提出了基于ORB(ORiented Brief)算法检测特征点，并采用最小凸包检测感兴趣区域的方法。首先，采用ORB算法提取出图像中的特征点，然后从中挑选出效果良好的点对图像进行描述，最后采用最小凸包算法检测出感兴趣区域。在和其它算法的检测速度对比和复杂情环境况下的检测实验结果表明，ORB和最小凸包算法的结合在保证的检测精度基础上提高了检测速度。

A Study on the Detection of ROI based on ORB and Minimum Convex Hull

With the development of science and technology, how to retrieve the region of interest(ROI) under complicated background and improve the efficiency of retrieval has become a pressing problem. To address this problem, this paper introduced a novel ROI retrieval method based on ORB and Minimum Convex Hull. Firstly, using the ORB algorithm to extracted the feature points in the image. Secondly, described the image by picking the excellent points from it. Finally, using the Minimum Convex Hull algorithm to extracted the ROI. Experimental results on detection speed and accuracy of the complex environment conditions show that ORB combined with Minimum Convex Hull algorithm can increase the detecting speed and guarantee detecting accuracy.