Recognition, Based On The Characteristics Of The Gait Of The Body Width

Gait recognition is a new technology of biometric authentication which absorbs more and more researchers to concern it in recent years; it recognises people by their way of walking. Similar to face recognition, it has to do two problems: identification and testing in application areas. In recognition, we will search the gait in the database to match the gait of unidentified. In testing, it is necessary to make a decision whether to accept or reject the assumed identity according to the algorithm of gait recognition. As a bio-authentication techniques, gait recognition have the unique advantages which other biological authentication technology doesn't have, such as recognise people far away, in low video quality, also, it is not easy to hide gait.Gait recognition analysis of human from image sequences involving humans usually includes gait motion segment, feature extract, and recognition. This interest is driven by a wide spectrum of promising applications in many areas such as virtual reality, smart surveillance and perceptual user interface. Gait analysis aims at attempting to detect, track and identify people, and more generally, to understand human behaviors. This paper pays more attention on gait detection, especially on the preprocessing of gait, feature extraction and classification. We also generalize the research state and used methods in and out.First, we provide a brief survey of recent developments of vision-based human motion analysis, and outline the theory and methods and the status and progress of the research of gait.Second, we detect the moving gait from each image sequence. In this paper we use the gait database which each sequence has only one person walk in the same background. Difference methods are used to extract the surveillance of moving gait. then, the outer contour of the silhouette is projected along the width of the body to obtain Width Projection Vectors, which are then translated into one dimensional vector as the gait feature.Then, eigenspace transformation based on PCA is applied to time-varying project vectors derived from a sequence of silhouette images to reduce the dimensionality of the input feature space. It will cause huge data and hard to deal with the identification If we use width vector directly as input. Because the small principal components involve important nonlinear feature, if we don't use it, we will lose a lot of important information. But, if we use all of the principal components, the system will hard to run. In the paper, compared the number of twenty, thirty and forty as the number of principal component, we chose thirty as the ideal one and get better result.Finally, gait recognition is performed by Support Vector Machine (SVM).SVM is the best theory and identification method in small sample statistic estimation and pre-estimate.In this paper, we separate the data set into two parts, one is used as training set, another is used as testing set. And we use RBF as kernel function to test our result.This paper uses the width vector as the gait feature, and combing the PCA and SVM to recognise gait. It implicitly captures the static silhouette and temporal characteristics of gait. And it has simple principle and is easy to realize. Experimental results on gait sequences demonstrate that the proposed algorithm has an encouraging recognition performance with relatively lower computational cost. Result proves that the method is very effective.