一种改进的梯度直方图远红外行人检测方法  

A far infrared detection recognition description method based on improved gradient histogram

作　　者：周红志[1] 刘祺[2] 

机构地区：[1]阜阳师范学院信息工程学院,阜阳236041 [2]华南理工大学计算机科学与工程学院,广州510640

出　　处：《四川大学学报（自然科学版）》2016年第5期1018-1026,共9页Journal of Sichuan University(Natural Science Edition)

基　　金：国家自然科学基金(61171141);安徽省自然科学重点研究项目(KJ2016A556)

摘　　要：在远红外行人检测领域,实时性以及鲁棒性如何提高,这一问题一直得不到有效的解决.笔者以感兴趣区域ROIs为切入点,在图像预处理阶段充分应用像素梯度相关知识,总结出有效的垂直投影方法.该方法原理为:搜集图像梯度信息,采用初定位方法对垂直方向上的条状区域展开处理并获取包括行人在内的图像,采取局部双阈值图像分割算法获得相对较为可靠性ROIs,避免对整幅输入图像进行搜索.检测过程中,充分发挥全局信息作用,采用以金字塔熵加权的梯度直方图方法,针对远红外行人进行描述.本方法引入了支持向量机算法的识别技术.文章最后针对该方法进行了实验检验.实验结果显示,该方法大大缩短了ROIs提取过程的搜索耗时,而且可以对一些以背景目标为主的候选区域进行抑制.For the real-time and robustness of far-infrared pedestrian detection remains a very challenging problem, the paper presents a vertical gradient projection method based on the pixels from ROIs （Re- gions of Interest） extracted angle. This method utilizes the image gradient information to conduct early positioning for the vertical banded image area which may contain pedestrians, then forms relatively relia- ble ROIs by partial dual-threshold image segmentation algorithm inside obtained image. In the pedestrian detection process, it combines global information and gradient histogram method based on pyramid weighted en- tropy to describe the far-infrared pedestrian, at the same time assist in SVM algorithm to complete the identifica- tion process. Experiments show that the method can improve the search efficiency of ROIs extraction stage, and suppress portion of the candidate region included in the background target.

关 键 词：行人检测 ROIs 像素梯度 双阈值 梯度直方图 SVM算法 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]