HOG/HOF Tensor Divergence Feature Extraction System based on HoG and HOF for video obejct action classification

摘要

The present invention relates to a method for extracting feature information for classifying object behavior from video, the method comprising the steps of: calculating a gradient vector and an optical optical flow vector for a selected key point for image frames; Obtaining a tensor product of the optical flow vector; and calculating a tensor divergence for the tensor product to reduce the dimension thereof, and determining the calculated tensor divergence as a feature vector for motion classification. According to the feature information extraction method and feature extractor for classifying the video object behavior, HOG and HOF are characteristics thereof, but they can reflect changes in space and time, It does not increase the amount of computation but classifies the behavior within the video, it provides the advantage that it can improve performance.;