计算机技术与发展2017，Vol.27Issue(1)：48-52,5.DOI:10.3969/j.issn.1673-629X.2017.01.011
复杂环境下基于视觉显著性的人脸目标检测
Face Target Detection of Visual Saliency in Complex Environment
摘要
Abstract
At present,the classification accuracy of face recognition in complex environment is not satisfied because of the background clutter and other factors. To solve this problem,a face detection algorithm based on visual saliency has been proposed,in which graph based visual salient algorithm is employed to extract salient maps of face region in complex environment and then threshold segmentation and morphological operations is run to get the binary face marks and the centroid is taken as center as well as the minimum distance of the region edge and centroid as side length to crop the accurate area of target face to achieve the goal of face detection in complex environ-ment. Results of experiments on the LFW image database show that the proposed algorithm can accurately fulfill“matting” tasks and a-chieve good results in face detection,the process of which needs no artificial participation,and can effectively exclude the interference of pell-mell background besides having improved the detection rate,realizing the unsupervised face detection,providing a theoretical foun-dation for intelligent research of face recognition.关键词
复杂环境/基于图论的视觉显著性算法/视觉显著性/人脸检测Key words
complex environment/GBVS/visual saliency/face detection分类
信息技术与安全科学引用本文复制引用
陈凡,童莹,曹雪虹..复杂环境下基于视觉显著性的人脸目标检测[J].计算机技术与发展,2017,27(1):48-52,5.基金项目
国家自然科学基金资助项目(61471162) （61471162）
江苏省自然科学基金(BK20141389) （BK20141389）
南京工程学院科研基金(QKJA201304) （QKJA201304）
