Realization Of Parallel Computing Based On Deterministic Learning In Gait Recognition

With the development of information technology, the computer can be used to serve usmore quickly and effectively. Especially with some high orders and large quantity calculationproblems, the performance of computer can directly influence the way of solving problems.At present, the level of computer processing has not been limited to depend on expansion inphysical memory and increase of processor speed, various types of parallel processing systemhas been applied already. So, the application of parallel computing can improve performanceof the computer by a big margin without the progress of processor ability. The principle ofparallel computing is to assign one task into different processes, and the processes can handlethe degraded small tasks at the same time. Thus, the speed of calculation is accelerated andthe efficiency of problem solving is improved. It will combine with many calculationproblems very well.In the premise of remote and disguised recognition, gait recognition technology mayrecognize personnel identity according to the human body gait characteristics. It’s veryessential in today’s life, because there exists more and more occasions in demand of IDrecognition. Now, the gait of recognition technology is based on the characteristics of thebehavior of the individual characteristics which includes image sequence preprocessing,period testing and feature extraction, and then the characteristics can be identified. Becausethe behavior of the human body is very complex, when modeling contain many unknown part.Because the behavior of the human body is very complex, there are many unknown partswhich can not be modeled precisely. The deterministic learning theory can solve this problemvery well. The deterministic learning theory which based on RBF neural network canapproximate the local system dynamics accurately in the unknown dynamic environment. Thedynamic gait characteristics which have been identified are stored in constant neural network,then the characteristics of the gait can be recognized quickly and accurately with this patternstorage.This paper mainly introduced the practical application of deterministic learning theory inthe gait characteristics identification. In order to speed up the identification process, improverecognition efficiency, the parallel computing is used in the gait recognition. In parallel hardware platform, the parallel recognition algorithms were instructed, and the parallelprogram was realized in multi-core system. At the end of the process, we analyze theperformance of the program. To reveal the procedure directly and conveniently, the GUI ofgait characteristics abstraction and recognition system is established by C++based on MFCframe and matlab GUI respectively.