基于视频流的快速行人检测技术的分析-analysis of fast pedestrian detection technology based on video stream.docxVIP

重庆大学硕士学位论文 英文摘要  PAGE 6 results prove its efficiency. Experimental results demonstrate that our proposed pedestrian detection system can significantly improve the detection speed without reduce the detection accuracy to both the video and the image based pedestrian detection with comparison to the traditional HOG based method. The complete detection system includes the post processing is specified in this thesis which may benefit to some object detection based engineering works. Key words: Pedestrian detection, Gaussian Mixture Model (GMM), PCA-HOG, Mean Shift, Support Vector Machine (SVM)