Tied multi-rubber band model for camera distance, shape and head movement robust facial recognition

Facial Capturing cannot ensure the stable head and fix the head position as well as accurate camera distance. This variation degrades the face recognition accuracy. For improving the recognition a true estimator is required to cover the camera distance and the head movement impact. This paper presents a rubber characteristics transformation to generate manifold camera distance and facial shape mapping. The work tied three rubbers to achieve estimated facial variation under shape and distance aspects. Rubber length and relative size ratio mapping are applied managing the face over the database. First rubber is placed between the camera and the foremost facial point at perpendicular. Here distance and size ratio variation is done for facial mapping. Second tied rubber is on the person body up to neck for same size level variation. Third rubber tied to represent the head itself with rotational and shape variation. A ring specific change variation is applied head map. All possible shapes are fused to generate the true head image for mapping. These three rubbers collectively provided shapes, direction and distance robust facial recognition. The comparative experimentation is provided against PCA, LDA and PCA-LDA methods for different sample sets taken from Aberdeen Datasets. The experimental results show that the presented work model provided the improvement over the recognition for more than 30%. The experimentation is applied on the individual variation and mix variations sample sets.