NCC-Template matching using Adaptive Statistical Fast LTP Features Makeup Faces

Abstract

Due to excessive use of low cost imaging devices the worldwide imaging data increases exponentially in recent times. Therefore template matching is widely used for authentication purpose for security and person identification and copyright protection. In real time uses complexity of template matching must be low but its efficiency must be good enough. Many complex or specific template matching methods based on local or global feature excretion were designed by researchers in past. The paper presented a simple CBIR based efficient template matching method. Method is a combination of statistical computing and LTP based texture feature extraction used for template matching. Method is designed for person identification to match the quarry face with template data base images based on statistical properties like face absolute mean differences and normalized correlation in 2D. The best RGB component is adopted with statistical entropy analysis for calculating the texture features. Method assumes that quarry and template images are of same size else resize them. The RGB component with maximum information is adopted for matching. The higher order efficient fast LTP ternary features are calculated for matching the texture patterns of the input images. The upper LTP features are matched using the normalized cross correlation based matching. The performance is tested for the face image data sets for face matching application for the color makeup images. Method works efficiently for different size and cases of the Face images