Image Contrast Enhancment Using Feed Forward Network

Abstract

Histogram Equalization (HE) is a popular, simple, fast and effective technique for improving the gray image quality. Contrast enhancement was very popular method but it was not able to preserve the brightness of image. Image Dependent Brightness Preserving Histogram Equalization (IDBPHE) technique improve the contrast as well as preserve the brightness of a gray image. Image features Peak Signal to Noise Ratio (PSNR) and Absolute Mean Brightness Error (AMBE) are the parameters to measure the improvement in a gray image after applying the algorithm. Unsupervised learning algorithm is an important method to extract the features of neural network. We propose an algorithm in which we extract the features of an image by unsupervised learning. After apply unsupervised algorithm on the image the PSNR and AMBE features are improved.