Streaming Video-based Gait Recognition

In ambient intelligence, biometrics identification has been a prevalent research field. Gait recognition, the second biometrics identification technology based on motion vision, is used to signify the identity of individuals in image sequences "by the way they walk ".Compared with other biometrics, such as fingerprint, face and iris, gait is unobtrusive and does not require to contact, so gait recognition may be performed at a distance surreptitiously. This paper, in both the theoretical and the practical perspective, probes into gait recognition with the videos as input. The main The main research and innovation of this paper are summarized as follows:1. Gait features demands for simple, effective and easy to classify, this paper presents a gait recognition algorithm, based on the video and integration of the different features of gait. This method can apply to a small scale. This algorithm also lays the foundation for further identification in large scale with a more precise method.2. Currently reported in the literature gait recognition algorithm, can be divided into basically based on the static shape feature and kinetic feature. This paper presents a method based on the combination of static shape features and dynamic movement features of gait, and make full use of the gait cycle in which a change in the shape of the movement and its own internal model changes.3. Based on the fact that the side of the body-image is symmetry, this paper Proposed a gait recognition algorithm based on human symmetry. We make an analysis of contrast between two feature choices, the front side and the back side of the body contour to the body symmetry axis distance vectors.4. Proposed a method based on integration of multiple features of both gait recognition algorithm. The first algorithm uses summation strategy, integration of the front side of the body contour to the body center of gravity distance vector, and the back side of the body contour to the body center of gravity of the distance vector and then for gait recognition, the second fusion algorithm using quadrature strategy, integration of the human side of the morphological model contour gait recognition. Through the integration of multi-feature recognition, the recognition rate than the algorithms used to identify a single characteristic gait recognition rate has improved.5. In the above study, on the basis of this paper, the proposed algorithm in a variety of conditions under the experimental verification. We set up our own data base, the algorithm in this paper for more comprehensive testing, therefore, the evaluation of this algorithm in a simple environment is basically objective and accurate.