Authors : Viet-Ha Ho, Duc-Hoang Vo, Van-Sy Ngo and Huu-Hung Huynh
Abstract: This study presents a method of person identification working on a sequence of 2D silhouettes. Inspired by the state-of-the-art gait signature gait energy image, we integrate a non-linear operation into the stage of feature extraction in order to enhance the obtained gait signature. Unlike related works, our method does not require the step of gait cycle separation. The accuracy of our approach was demonstrated by experiments on the CASIA-A gait dataset and was comparable with many related studies.
Viet-Ha Ho, Duc-Hoang Vo, Van-Sy Ngo and Huu-Hung Huynh, 2019. Person Identification Based on Euclidean Distance Transform. Journal of Engineering and Applied Sciences, 14: 4312-4316.