Content Retrieval Algorithm Based on Improved HOG

Histograms of Oriented Gradients (HOG) feature has been successfully used in pedestrian detection and achieves high accuracy. This paper introduces a content retrieval algorithm based on improved HOG. The method has two steps which are adjusting the HOG structure by scanning the image with a sliding HOG window and reducing feature dimension by principle component analysis (PCA) technique. The experimental results show that: the precision rate of this method has improved significantly compared with the method of transforming the size of images to calculate HOG feature and the method of extracting color feature.