说明：  这是一个完整版，包含调用文件。整个视频可以用在不同尺度和位置计算得来的特征集进行表示。Hog3D描述子由Alexander Klaser Marcin Marszał ek Cordelia Schmid等人提出的，将HOG的概念扩展，由静态图像特征提取扩展到视频序列的特征提取，在视频序列中的行人检测取得了较好的成果。
(This is a complete version, including calls to the file. The video can be used in different scale and location characteristics were calculated set of representation. The Hog3D descriptor proposed by Alexander Klaser Marcin Marsza EK Cordelia Schmid et al, will the concept of HOG is extended static image feature extraction, extended to extract features of video sequences, pedestrian detection in video sequence and achieved good results.)