The Gait Recognition Technology Research And Implementation In Video Mining

With the development of the society, the progress of science and technology, various fieldsmore and more high to the requirement of safety, for human action recognition technologyrequirements of the surveillance video is becoming more and more high, such as gait recognition,face recognition, and iris recognition. By the human eye to find key information from these largeamounts of video video not only waste a lot of time and also waste a lot of unnecessary energy,thus was born the video mining technology. Because the gait characteristics hidden, non-contactdistance and not easy to pretend, in the field of intelligent video monitoring, have moreadvantages than other characteristics, so this paper USES the gait as a feature of surveillancevideo recognition.This article research is with a fixed camera surveillance video, need to use camera to videosegmentation are divided into several or a dozen camera (according to the size of the video), inthis paper, using the histogram partitioning method, make the back of the video mining and savestime.Recognition is using hu moment to calculate the gait energy image. Hu moment can beretained the most for good information and the minimum errors, so it uses hu moments in thearticle. Use the matrix sequence obtained to identify by SVM and then compare similauity. Theshots are divided into key shots and blank shots, and the key shots are divided into target shotsand detection shots. The target shots are confirmable, and the detection shots are compared withthe target shots to find the corresponding detection shots of similarity large rate.Corresponding the detection shots contains many information where is needed keyframeextraction. In the link of the key frame extraction, it uses Euclidean distance, mean, variancecharacteristics on the difference between the frames to calculate and screens out the key frameby using the extreme value and intermediate value comparison. The accuracy rate of this methodimproves25%than previous ones, but the disadvantage is that the time complexity improveds15%.