Pedestrian detection method under low resolution and based on sparse representation processing

Hu Ruimin,Fang Wenhua,Liang Chao,Nan Yuanyuan,Wang Xiao,Pan Yifei,Chen Jun

2014-01-01

Abstract:The invention provides a pedestrian detection method under a low resolution and based on sparse representation processing. The pedestrian detection method comprises the steps of obtaining a high- and low-resolution complete dictionaries, performing linear representation on low-resolution surveillance video images by use of the low-resolution dictionary, and a linear representation coefficient by the high-resolution dictionary so as to synthesize high-resolution images to be tested corresponding to the low-resolution surveillance video images, next, extracting HOG (Histogram of Oriented Gradient) characteristic vectors from the synthesized high-resolution images to be tested, and then carrying out classification and identification by use of an SVM (Support Vector Machine) classifier to obtain the classification result of the high-resolution images to be tested, thereby determining whether the low-resolution surveillance video images contain people. The pedestrian detection method is capable of obtaining more accurate pedestrian search results by converting the low-resolution surveillance images into the high-resolution images before detection.