基于形状上下文的驾驶室值班人员检测算法

Detection algorithm based on the shape context for cab duty officer

提出了一种基于形状上下文的人脸模板匹配算法。首先,对待检测图像进行椭圆状目标检测,标记出所有的"候选头部区域";然后,根据形状上下文算法将"候选头部区域"分割出来,与人脸模板分别匹配,计算出其对数极坐标直方图,得到匹配图像的Cs值。最后,判断Cs值大小来确定模板与待测目标是否匹配,输出最终标记结果。该算法具有二维目标不变性的特点,能够为目标正确标记提供可靠依据。

A method for human face shape matching algorithm based on shape context.First,detect the image where has the oval shape,and market them as the ＂candidate head area＂.Then,use the shape context algorithm to cut up the ＂candidate head area＂ from the image,put them to match with face template,and calculate the log polar histogram,then,get the Cs value of the image.Finally,judge the size of Cs value to determine if the template and the target to be detected are matched,output the final mark results.The algorithm has a feature of two-dimensional invariance,it will be able to provide a reliable basis for target marking.