MULTIVIEW HUMAN GAIT ANALYSIS USING THE FIRST AND THIRD PERSON DATA

Abstract

The gait of a person is often used as a biometric tool to identify or classify a person based on gender and age. Because of its ability to track a person from afar, gait recognition has found use in a variety of fields, including forensics, surveillance, and health monitoring departments. Biometric systems is a rapidly emerging area that necessitates the development of new methods to address problems that have plagued previous approaches. Human gait is a less-explored region in the field of biometrics. In this Project, two different types of gait datasets have been reported and presented. The FP (First Person) data carrying the camera motion collected from the movement of the volunteer's body and the TP (Third Person) data captured from a distant view were recorded at the same time. A total of 24 subjects (15 males and 9 females) are included in the dataset. The discussion is extended to include a comparison of the results obtained using TP and FP data. This report also provides an extensive survey of the stages involved in the framework of gait recognition by analysing the different methods used in each stage along with the description of the feature extraction process and the state-of-the-art techniques used in appearance-based and human-pose-based methods. Moreover, a brief comparative description on the recent data reduction or feature selection methods has been provided. Furthermore, it will provide a first-hand knowledge about the public datasets that is motion capture databases and the datasets simply used for human gait recognition. In comparison to other biometric methods, gait recognition has a lot of potential for future work, Researchers working in the fields of biometrics, human pose estimation, monitoring, human gait recognition and analysis will benefit from the review given in the survey.