Comparative statistical analysis of the quality of image enhancement techniques

Abstract

Image Enhancement aims at processing an image in such a manner that the resultant is more appropriate than original image for any specific application. This article focuses on image enhancement, with particular reference to different image fusion and spatial filtering techniques. Statistical analysis of image quality measures was used for evaluating the quality of enhanced images. During the work, image enhancement was carried out in two steps. Firstly, to enhance the spatial resolution, the effectiveness of eight diverse image fusion techniques (brovey, wavelet, IHS, HPF, ehlers, PCT, subtractive, and Hyperspherical Colour Space (HCS)) was examined. These fused images were evaluated using different objective image quality measures like CC, entropy, NLSE, AG, MAD, ERGAS, SD, RASE, PSNR and MAPE. These measures helped in determining the preservation of spectral and spatial integrity in the fused images. Secondly, the enhanced fused image of the best quality was subjected to six different filtering techniques (high pass filter, low pass filter, adaptive, edge enhance, sharpen, and focal analysis). Based on the statistical results of the image quality measures, the image using HCS fusion method followed by low pass filtering or focal analysis is selected as a recommendation for enhancement of an original Landsat image for any specific application.