An Artificial Neural Network based Adaptive Histogram Equalization Algorithm for Enhancement of Low Contrast Images

The implementation of an image contrast enhancement algorithm along with artificial intelligence techniques can have various applications besides modern photography. It basically ameliorates the quality of low contrast images. The main focus of this research is developing a new image contrast enhancement method that combines the concept of artificial intelligence and histogram equalization techniques to provide a contrast distribution for the low contrast images by utilizing the classifier to prevent data loss from images. In this research an ANN based AHE algorithm for enhancement of low contrast images is proposed. The main objectives of this research is to study the existing digital image contrast enhancement techniques to find out the exact problems and to classify the level of contrast in a digital image as low or high, so as to ascertain whether enhancement is required or not. The concept of ANN with AHE is used here to find out the contrast level of the image before processing for contrast enhancement. For validation of the proposed ANN-AHE algorithm, a comparison with the existing techniques are performed on the behalf of performance parameters such as PSNR, MSE, Entropy, QI, QRCM, CQE, SSIM and Computational Time. The simulation of the proposed model is performed in MATLAB 2016a with the help of image processing and artificial neural network toolbox.

I. Introduction

Digital Image Processing (DIP) plays an important role in different application areas such as analyses of the image, forgery detection, video surveillance, medical science imaging, and Satellite or hyper-structural imaging. The major concerns with the acquired images are their noise and low contrast. Image contrast decides the quality of an image, so lots of analyses are already presented by the researcher in the areas of the image contrast enhancement. In the DIP, contrast enhancement performs the adjustment on the darker or lighter portions of the image. This basically brings out the hidden features in the image. The area of DIP has been categorized into two different domains such as spatial and frequency domains to perform the contrast enhancement. The proposed research even though focuses mainly on the spatial domain, it discusses method related to the frequency domain as well. As the method proposed in this research work is related to the Histogram Equalization which improves the contrast of the image.