Pedestrain Detection Based on a Gabor Weber Local Descriptor
Download as PDF
DOI: 10.23977/CNCI2020058
Corresponding Author
Guoyun Lian
ABSTRACT
Pedestrian detection has attracted more and more attention in recent years. In this paper, a novel pedestrian detection method based on Gabor Weber Local Descriptor (GWLD) was proposed. According to the pedestrian's characteristics, the sliding window technique was adopted. Firstly, Gabor transforms were performed on the sliding window, and then the Weber Local Descriptor (WLD) was extracted over the average Gabor feature maps. Finally, the GWLD histogram was constructed to character the sliding window. Experimental results on INRIA pedestrian dataset and Daimler Chrysler (DC) pedestrian dataset validate that the effectiveness of the proposed GWLD detector. Comparing with the other pedestrian detection methods, the proposed GWLD method performs better.
KEYWORDS
Pedestrian detection; Gabor transform; weber local descriptor