摘要
针对复杂场景中背景复杂、目标周围噪声多及目标只占图像中较小部分而难于检测的问题,提出一种新的基于局部轮廓特征的检测目标方法.该方法首先利用改进的全局概率边界算法(Globalized probability of boundary,gPb)算法提取图像的轮廓,然后应用最大类间方差法(Otsu)进行自动阈值处理得到图像的显著性轮廓;再提取显著性轮廓的k邻近大致直线轮廓段(k connected roughly straight contour segments,kAS),并以kAS作为局部特征,用于复杂场景中的目标检测.该算法结合gPb算法和Otsu提取轮廓的显著性轮廓,去除了目标附近的大量噪声边界,有效地提高了检测效率.同时,在检测阶段,测试集与训练集中提取的不相关特征数目也得到较大减少,从而提高了检测的精度.多组实验结果均表明本文方法的有效性.
It is difficult to detect objects in complex scene in which more noise is around the object or the object is only a small portion of the image. In order to solve the problem, a new object detection algorithm based on local contour features is proposed in this paper. Firstly, an improved gPb （globalized probability of boundary） Mgorithm is used to extract the outline of the image. Then the Otsu for automatic threshold processing is applied to obtain the significant contour. Next, k connected roughly straight contour segments （k adjacent segments, kAS） are extracted and used as a local feature for object detection in complex scenes. The algorithm combines gPb algorithm and Otsu to extract significant contour, thus it can remove much noise around the object boundary, and effectively improve the detection efficiency as well. Meanwhile, in the detection phase, the numbers of irrelevant features in the test set and the training set are largely reduced, therefore the detection accuracy is improved. Multiple sets of experimental results demonstrate the effectiveness of this method.
出处
《自动化学报》
EI
CSCD
北大核心
2014年第10期2346-2355,共10页
Acta Automatica Sinica
基金
国家自然科学基金(61063030
61263046
61165011)资助~~
关键词
轮廓提取
局部轮廓特征
阈值处理
目标检测
Contour extraction, local contour features, threshold processing, object detection
