The Design Of Gait Rapid Recognition System Based On Deterministic Learning

Recognition by gait is a new and attractive research field for biometrics recognitiontechnology. Its aim is to recognize people or detect physiological, pathological and mentalcharacteristics by their walking style. The use of gait as a biometric for human identificationis promising. The technique of gait recognition, as an attractive research area of biomedicalinformation detection, attracts more and more attention. It consists of three phases such asfeature extraction, feature training and recognition. In view of the gait recognition system haspotential applications in areas such as visual surveillance, the paper has carried on intensivestudy on the realization of gait recognition system.Deterministic learning is a new theory applying to local accurate identification fornonlinear system. The deterministic learning-based approach for human gait recognitionconsists of two phases: a training phase and a test (recognition) phase. Locally-accurateidentification of the gait system dynamics is achieved by using radial basis function (RBF)neural net-works (NNs) through deterministic learning. The obtained knowledge of theapproximated gait system dynamics is stored in constant RBF networks. Hence, time-varyinggait dynamical patterns can be effectively represented by the locally accurate NNapproximations of system dynamics, and this representation is time-invariant. A gait signatureis then derived from the extracted gait system dynamics along the phase portrait of jointangles versus angular velocities. A bank of estimators are constructed using constant RBFnet-works to represent the training gait patterns. In the test phase, by comparing the set ofestimators with the test gait pattern, a set of recognition errors are generated, and the averageL1norms of the errors are taken as the similarity measure between the dynamics of thetraining gait patterns and the dynamics of the test gait pattern. Therefore, the test gait patternsimilar to one of the training gait patterns can be rapidly recognized according to the smallesterror principle.This dissertation has carried on the exploration and research of gait recognition system inpractical application, based on the related theory of existing gait feature extraction algorithmand human gait recognition approach via deterministic learning, the design of rapid gaitrecognition system will be further studied in the dissertation. The main contribution andinnovation of this dissertation are summarized as follows:(1) A LABVIEW-based human gaitrecognition off-line system is designed. The effectiveness and rapidness of the gaitrecognition approach under multi-pattern is verified on the NLPR gait database. Therecognition speed is improved by using the parallel programming mode of LABVIEW and the multi-core CPU of A840r-H Dawning Server.(2) A GPU-based human gait recognition realtime online system is designed. To solve the problem of large scale computation in practicalapplication for real-time fast gait recognition-based on large scale database, we implementedthe gait recognition algorithm parallelizing on the JACKET platform. Specific experimentsare performed under the environment constructed by ourselves, the effectives of the gaitrecognition system based on MATLAB JACKET platform real-time online is verified. Theinnovation on gait recognition system realization has practical application value.