Gait Recognition-A Novel Approach to Quality Improvement in Human Silhouettes
This paper proposes a robust algorithm for the quality improvement in human silhouettes, to improve the gait recognition percentage of a person. In silhouette based gait recognition approach, the presence of incomplete and noisy silhouettes has a direct impact on recognition performance. Using blob detection, initially the incomplete silhouettes are identified. Fusion of frame difference energy image with dominant energy image of a silhouette along with a morphological filter output, preserve the kinetic and kinematic information to make incomplete silhouette into a high quality and a complete silhouette. The results prove that the resultant silhouettes are well suited for human gait recognition algorithm with improved variance. The silhouette database is taken from CASIA database. (Institute of Automation, Chinese Academy of Sciences).