微型机与应用2016，Vol.35Issue(21)：14-15,19,3.DOI:10.19358/j.issn.1674-7720.2016.21.004
HOGG:基于Gabor变换与HOG特征的人体检测
HOGG: Gabor and HOG-based human detection
摘要
Abstract
Aiming at the multi-scale characteristic of human body in visual surveillance, the multi-scale orientation features (HOGG) for pedestrian detection are put forward in this paper.Firstly, the multi-orientation information corresponding to human body image are extracted, and then fusing the original Gabor features of the same scale in different directions into a single feature in order to reduce the dimension of the features.Then histograms of oriented gradients of each fused image is extracted and combined as human image features.Support vector machine (SVM) is adopted to sort the features, and experimental results on CAVIAR database show that the proposed method has a better performance for human body target detection.关键词
人体检测/Gabor变换/分块直方图/多尺度Key words
human detection/Gabor transform/block histogram/multi-scale分类
计算机与自动化引用本文复制引用
范国娟,范国卿,柳絮青..HOGG:基于Gabor变换与HOG特征的人体检测[J].微型机与应用,2016,35(21):14-15,19,3.基金项目
山东省高等学校科技计划项目(J15LN53) （J15LN53）
