Person identification based on gait using dynamic body parameters

Gait as a behavioural biometrics has been the subject of recent investigations. One of the unique advantages of human gait is that it can be perceived from a distance. A varied range of research has been undertaken within the field of gait recognition. A gait describes the manner of a person´s walking. It can be acquired at a distance and if necessary without consent or knowledge of the subject. Human gait representation can be roughly divided into two categories. One is model-based gait approach and other is model free gait approach. A human body feature that contributes more to an automatic gait classification is subdivided into two i.e. static (body shape) or dynamic (the movement of legs and arms). In this research work, we have considered dynamic features of human body for gait recognition. In our proposed research work, we have considered two features of human body i.e hand and feet for gait recognition. Second feature feet is subdivided into two i.e toe and heel. Both left and right legs toe and heel are considered. We follow an approach of parametric line equation for formulating two triangles between these features i.e first triangle is formed between hand and toe of both legs(right and left) and second triangle is formed between same hand and heel of both legs(right and left). After triangles formation we have find two intersecting points between these triangles and angles at these intersecting points. We have then calculated mean value of angles and intersecting points of an individual subject gait cycle frames and match mean value with database for recognition. We have designed a gait system using Matlab R2007b to accomplish this research work. We have worked on gray level images. Evaluation about the proposed gait recognition method is given according to experiments on CASIA gait database and the experimental results demonstrate the encouraging performance.