Applying gamma and histogram equalization algorithms for improving system-performance of face recognition-based CNN

Abstract EN

In the last few years, many applications have viewed great development, such as smart city applications, social media, smartphones, security systems, etc.

in most of these applications, facial recognition played a major role.

the work of these applications begins by locating the face within the image and then recognizing the face.

the circumstances surrounding the person at the moment of taking the picture greatly affect the accuracy of these applications, especially the inappropriate lighting.

therefore, the stage of preparing the images is very important in the work.

to solve this problem, we proposed a system that combines the use of gamma and histogram equalization algorithm (HE) to improve the images before starting to detect the face using the Viola-Jones.

then extract the facial features and identify the person using convolutional neural networks.

the proposed system achieved a very small error rate and an accuracy during training that reached 100%.