Feature integration for frontal gait recognition through contour image analysis

Feature integration for frontal gait recognition through contour image analysis

연구 분야: Artificial Intelligence

학회: Signal, Image and Video Processing

Biometric technology is advancing with gait recognition, which analyzes walking patterns to identify people. This pattern is derived without the direct participation of individuals, from a distance. Frontal gait data is highly valuable in confined spaces like narrow corridors, which is common in most buildings. Within this scope, this study introduces a successful approach to identify individuals in frontal-view gait sequences. By utilizing contour image and vertices, the proposed method obtains three differentiating feature vectors from the Gait Energy Image (GEI). Its efficient capture of spatial dynamics leads to improved gait recognition performance. The proposed approach’s effectiveness was evaluated using the widely used gait datasets such as CMU MoBo, CASIA A, and CASIA B. Through the experiments, it was proven that the proposed approach delivers promising outcomes and performs better than certain state-of-the-art approaches in recognition.