Survey: Gait Recognition Based on Image Energy

Gait, a biometric trait, is a highly valuable indicator that enables the identification of an individual from a significant distance and without requiring direct physical contact. Due to its unique attributes that are absent in other commonly used identifiers such as fingerprints and iris recognition, gait recognition has become increasingly prevalent and widely adopted in various areas where it is feasible to utilize video reconnaissance frameworks. As computer vision techniques have advanced, there are now numerous methods for recognizing individuals through their movements in the video, utilizing both traditional biometric characteristics such as the human skeleton, its silhouette, and its changes during walking, as well as unique signals. This paper explores the use of gait energy extraction as a fundamental approach for discriminating between different gaits, an analysis of the classification algorithms employed, and a comparison of the methods based on accuracy, gait datasets, as well as their characteristics, issues, and challenges encountered in the development of gait recognition systems. and identified the prevailing methods employed by researchers as GEI (Gait Energy Image) and FDF(Frequency Domain Feature ). Followed by GENI (Gait Entropy Image), CGI (Chrono-Gait Image ), and GFI(Gait Flow Image ). Among the available datasets, CASIA and OU-ISIR as the most extensively used and comprehensive datasets in the field. Using convolutional neural networks, they demonstrated that CGI achieved the highest level of accuracy when applied to the CASC-B dataset.