摘要
已有的人体上半身检测方法一般是建立在行人站立或者是行走的平视图上,需要从多个角度对人体进行检测。为此提出基于有向直方图的人体上半身检测方法,该方法把人体上半身的梯度方向直方图统计信息作为检测目标的特征,提取检测样本,从而准确的对人体上半身进行检测,通过实验分析发现提出的方法不仅检测效率高且检测速度也相对较快。
The existing methods of detection of the upper body of the human body are generally based on the flat view of standing or walking, which requires the human body to be detected from multiple angles. Therefore offered based on a histogram of human body detection method, this method the human upper body gradient direction histogram statistics as a test, the characteristics of the target extraction test sample, thus accurate to test the human upper body, through the experiment analysis found that the proposed method not only inspection with high efficiency and speed is relatively fast.
引文
[1]孔月萍,刘霞,谢心谦,等.基于梯度方向直方图的人脸活体检测方法[J].激光与光电子学进展,2018,55(03):149-151.
[2]徐海洋,孔军,蒋敏,等.基于时空方向主成分直方图的人体行为识别[J].激光与光电子学进展,2018,55(06):201-203.
[3]郭永茂,周石博,高艾.基于方向梯度直方图的陨坑预检测算法[J].无线电工程,2018,48(06):478-483.