Face identification based on biological trait using infrared images after cold effect enhancement and sunglasses filtering

Abstract

This paper describes face identification using infrared images and eigenfaces after passing test face through cold effect enhancement and/or sunglasses filtering algorithms and handling facial hair through threshold. Eigenface technique after modification is used to define our eigenspace. Test image before going through the recognition process has to pass through a check to see whether it is a face image or not. The test face is passed through an algorithm to check and enhance if the person come from cold and then is projected to eigenspace to find the match. If match is not found then it is passed through another algorithm to check whether person has worn sunglasses and if so the image is enhanced in order to make recognition more efficient. Only one person of the test sets is recognized wrongly with the other but both of them belongs to the training set and it gives 100% accurate results for profile images.