Human Identification Using Gait

Nayak, Abhijit (2015) Human Identification Using Gait. MTech thesis.



Keeping in view the growing importance of biometric signatures in automated security and surveillance systems, human gait recognition provides a low-cost non-obtrusive method for reliable person identification and is a promising area for research. This work employs a gait recognition process with binary silhouette-based input images and Hidden Markov Model (HMM)-based classification. The performance of the recognition method depends significantly on the quality of the extracted binary silhouettes. In this work, a computationally low-cost fuzzy correlogram based method is employed for background subtraction. Even highly robust background subtraction and shadow elimination algorithms produce erroneous outputs at times with missing body portions, which consequently affect the recognition performance. Frame Difference Energy Image (FDEI) reconstruction is performed to alleviate the detrimental effect of improperly extracted silhouettes and to make the recognition method robust to partial incompleteness. Subsequently, features are extracted via two methods and fed to the HMM based classifier which uses Viterbi decoding and Baum-Welch algorithm to compute similarity scores and carry out identification. The direct method uses extracted wavelet features directly for classification while the indirect method maps the higher-dimensional features into a lower dimensional space by means of a Frame-to-Exemplar-Distance (FED) vector. The FED uses the distance measure between pre-determined exemplars and the feature vectors of the current frame as an identification criterion. This work achieves an overall sensitivity of 86.44 % and 71.39 % using the direct and indirect approaches respectively. Also, variation in recognition performance is observed with change in the viewing angle and N and optimal performance is obtained when the path of subject parallel to camera axis (viewing angle of 0 degree) and at N = 5. The maximum recognition accuracy levels of 86.44 % and 80.93 % with and without FDEI reconstruction respectively also demonstrate the significance of FDEI reconstruction step.

Item Type:Thesis (MTech)
Uncontrolled Keywords:Gait Recognition, Human Identification, Silhouettes, Surveillance, Biometric, Computer Vision
Subjects:Engineering and Technology > Electronics and Communication Engineering > Image Processing
Divisions: Engineering and Technology > Department of Electronics and Communication Engineering
ID Code:7888
Deposited By:Mr. Sanat Kumar Behera
Deposited On:16 Sep 2016 16:35
Last Modified:16 Sep 2016 16:35
Supervisor(s):Ari, S

Repository Staff Only: item control page