Pedestrian Detection in Infrared Images using Histograms of Oriented Gradients and Wavelet Entropy

J. Li and W. Gong (PRC)


pedestrian detection, infrared images, histograms of oriented gradients (HOG), dual-tree complex wavelet transform (DT CWT), support vector machine (SVM)


The paper proposes an efficient method for pedestrian detection in infrared images using histograms of oriented gradients (HOG) and dual-tree complex wavelet transform (DT CWT) based wavelet entropy features. It first locate the regions of interest (ROI) based on the high brightness property of the pedestrian pixels. Then the algorithm combines HOG and DT CWT-based wavelet entropy features to describe the ROI. Taking the combined features as input vector, it finally uses the support vector machine (SVM) classifier to classify and recognize the true pedestrian region. Experimental results using several infrared image databases demonstrate that the presented method is both effective and efficient.

Important Links:

Go Back