Gait Recognition Based On Image Processing Technology

Gait recognition is a new technique of biometrics recognition. Along withdiversification and high-tech of criminal means, the technology for human identifyis high required in safety area. The gait recognition has these advantages foruntouched, long distance and hard to be hidden, so many researchers who studyingbiometrics have been attracted recently.The arithmetic for gait recognition is been study in this thesis. First of all,current methods about gait recognition have been analyzed deeply, it include severalstages composed by target segmentation, feature extraction and pattern classification.Then the background image model has been established. The target has beensegmented by difference background with current image. The image is binarization.A new template is designed and the crack in image is eliminated by using ofmorphologic algorithm. The algorithms that projecting and normalized of imagetarget area is proposed. When all the algorithms are taken, we get the gait image thatthe most noise is filtered and the complexity of calculating is also decreased.It is studied that the Zernike invariant moments is used to extract gait features.Another algorithm for gait features extracted is proposed based on the area coveredby legs. Clustering samples under the monitor learning method and get the samples classification. All algorithms proposed above are simulated on computer and theresults are satisfactory.In the end, a system for gait image processing is established. The hardware andsoftware of the system is fulfilled. So, it can be uses in gait research in the future.