A Hybrid Face Image Contrast Enhancement Technique for Improved Face Recognition Accuracy

Abstract

Automatic Face recognition system struggles to recognize face images acquired at varying illumination conditions, facial expression, aging, and pose. The focus of this research work is to enhance the illumination affected face images, which subsequently results in improved face recognition accuracy. This paper presents a new contrast enhancement technique for face images. It is a hybrid contrast enhancement technique based on the combinatorial approach of Completely Overlapped Uniformly Decrementing Sub-block Histogram Equalization (COUDSHE) with various neighbourhood metric. COUDSHE concentrated to bring out a unique framework for localizing a global technique to adapt efficiently to local lightening conditions. The idea of using neighbourhood metric with histogram equalization results in enhancing the contrast of the image. Hence, the combination of COUDSHE with neighbourhood metrics results in enhanced images that are both adaptive to local brightness and good contrast enhancement. The application of our hybrid technique on the Extended Yale B Database has proved to improve the contrast of images and as a result, there is significant improvement in the recognition accuracy of the face recognition system implemented using Principal Component Analysis.

Handwritten Character Recognition Using Unsupervised Feature Selection and Multi Support Vector Machine Classifier

“…Calculate the 𝑚𝑖𝑛 peaks of variance ← 𝜇 1,2 × 𝐶𝐷𝐹 The second scenario is to compare several extractors of texture features including LBP, LTP, COUDSHE with Fuzzy [25], IALTP+ [26], the Proposed Method (MBALTP). For LBP, we implement the steps on a previous study [18] because the original LBP study uses different dataset, therefore the result cannot be compared directly.…”

“…The third method is COUDSHE (Completely Overlapped Uniformly Decrementing Sub-block Histogram Equalization) with Fuzzy [25]. The method was proposed by S. D. Ganesan and M. A. R. Mohammed.…”

Real Time Eyeball Movement Detection Based on Region Division and Midpoint Position

“…In this case, we used the Haar Cascade classifier method that has the ability to detect an object appropriately [13]. The detection process in the Haar Cascade was done by changing the RGB image into grayscale image and using integral calculation with some image enhancement technique, considering good image quality, can improve the detection process [14].…”

Analysis of Accuracy Rate of Distributed Virtual Reality Face Recognition Based on Personal Intelligent Terminal

“…However, due to the limitation of the comprehensive reasons of theory, practice, and general environment, the problem of face recognition did not attract the attention of academic circles until the middle of the 20th century. In recent years, with the rapid development of society and science technology, computer vision technology, and pattern recognition technology, facial recognition technology has gradually become a hot topic in the field of vision and recognition research [1,2]. Since the "9.11" incident in the United States, people have put forward higher requirements for information security and concealment, including how to break through the traditional facial inspection and recognition methods.…”