摘要
对红外图像中的人体检测问题进行了研究,提出一种快速、高效的检测算法。针对红外图像中人体图像的特点,通过方向投影确定了人体候选区域可能存在的位置,进而采用方向梯度直方图对候选目标进行了特征描述。将方向梯度直方图作为输入向量采用Fisher线性判别和贝叶斯分类器对候选目标进行了分类,以对候选目标中存在的人体进行检测。大量的实验结果表明所提出的算法是有效的。
The pedestrian detection problem was discussed, and a fast and efficient detection program was proposed. The regions of interest （ROI） in infrared image were located based on the high brightness property of the pedestrian pixels, and then histograms of oriented gradients （HOG） were adopted to describe the ROI. Taking HOG as input vector, the pedestrian region was detected through Fisher linear discriminant and Bayesian classifier. Experimental results demonstrate that the presented method is both effective and efficient.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2007年第19期4490-4494,共5页
Journal of System Simulation
基金
国家自然科学基金项目(60632050
60472060)
关键词
人体检测
方向梯度直方图
红外图像
FISHER线性判别
贝叶斯分类器
pedestrian detection
histograms of oriented gradients
infrared image
fisher linear discriminant
Bayesian classifier
