SATELLITE IMAGE CONTRAST ENHANCEMENT USING MODIFIED HISTOGRAM

Abstract

This project presents a study of various Histogram Equalisation based Contrast Enhancement (CE) techniques followed by the proposal of a novel CE algorithm for satellite and aerial images. The algorithm is referred as Contour Based Histogram Equalisation (CBHE). The algorithm presents a novel method to capture the structural property of an image using the contour of the image. The algorithm addresses the inherent drawbacks of HE viz. artefacts and saturation by decreasing the contribution of high probability pixels in the histogram and increasing that of low probability pixels. Finally the algorithm enhances the features of the image by adjusting the coefficients of DCT. The algorithm generates good contrast images with richer details over a varied set of images including satellite and aerial images. It is computationally comparable to HE, does not introduce noise and saturation, preserves characteristic shape of original image histogram and does not enhance an already high contrast image. It thus qualifies as an effective pre-processing step. CBHE is compared with the conventional and best of HE based techniques both quantitatively and visually. The quantitative analysis of the results is carried out using several standard measures like Discrete Entropy, Signal to Noise Ratio (PSNR), Measurement of Enhancement (EME), Average Mean brightness Error (AMBE), Gradient Magnitude Similarity Index (GMSD) and Structural Similarity Index (SSI) over varied datasets.