序列图像处理的快速人体检测方法

在序列图像中进行运动人体检测 ,一般都采用差图像算法 ,作者在应用尺度滤波器消除噪声的基础上 ,提出了将数学形态学的闭运算用于人体轮廓断裂处的连接填充。在此基础上得出人体轮廓的宽度直方图 ,并利用人体几何尺寸的固有比例对人体轮廓检测算法进行改进。实验结果验证了上述方法对噪声抑制、人体图像断裂处填充及人体识别的有效性

a MMethod for Fast human Detection Based on Sequence Image Processing

Difference image processing algorithm is adopted currently in detection motion human body in the sequence image. In this paper, on the basis of removal of noise using scale filter, a method,which filling the gaps of segmented human body using mathematical morphological close operation,is put forward. Sequentially, the width histogram of human body is gained. Under making use of the dimension of human, detection algorithm of human body has been improved. The experimental results demonstrate that the above algorithms is efficient for noise suppression, image gap filling and human body recognition.