Human Identification Research From A Distance Based On Gait Recognition

As the need of intelligent monitoring system in security sensitive occasions,gait recognition more and more plays an important role in human identification from a distance.Gait recognition system are composed of three parts,such as pretreatment,feature extraction,identification classification.At the beginning,in view of shortcoming that classical background abatement method can't examine the small part between moving object and the background,using method of double threshold binarization improve background subtraction method obtain the good test result.Next, using angle histogram method extract gait information.Theoretical analysis and experiment indicate that the contribution of different part of angle histogram are not the same for identification classification.Finally,euclidean distance were used as similarity metric.At the premise of the efficient gait information,the principle of the farthest distance between categories are used in identification classification.The correct classification rate is 78.5%.The experiment shows that the algorithm is effective.