Abstract
Face recognition (FR) system is one of the prominent research areas in pattern recognition and computer vision because of its major challenges. Few challenges in recognizing faces are blur, illumination, and varied expressions. Blur can be uniform and non-uniform. Most probably, blur occurs while taking photographs using cameras, mobile phones. Non-uniform blur happens in images taken using any handheld image devices. Recognizing or handling a blurred image in a face recognition system is generally hard. Most algorithms like blind deconvolution, deblurring, etc. are present and their problems have been discussed. To solve the issue of blur, Blur Robust Face recognition algorithm (BRFR) has been done. In Blur Robust Face Recognition, the Local Binary Pattern (LBP) features were extracted from the blurred test image. For each image in the set of training images, the value for optimal Transformation Spread Function (TSF) was found in order to form a transformed image. The LBP features were extracted from the transformed image. Both the LBP features of blurred test image and transformed training image were compared to find the closest match. Yale face database and JAFFE database were used for evaluation. By varying the Blur levels for the test image, the results for both the datasets were obtained.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call