Human gait recognition by micro-doppler analysis of the signal

Abstract

As a biometric, also described as behaviometrics, gait is playing an increasingly important role in identification applications in security and surveillance. Compared with physiological biometrics such as fingerprints or retina, it has behavioral characteristics which can be used to label or describe an individual according to their pattern of walk, theoretically supported by Micro-Doppler effect. In this report, an automatic gait recognition system is proposed and developed to achieve gender recognition and identity recognition. Various normalization and feature extraction methods are attempted and adopted to achieve a high accuracy. The final result presents an accuracy 86.94% for identity recognition. Concerns and issues raising from the usage of the system have been discussed and further recommendations are made for better performance of the system.