Quality assessment of pan-sharpening methods in high-resolution satellite images using radiometric and geometric index

  • PDF / 12,349,402 Bytes
  • 10 Pages / 595.276 x 790.866 pts Page_size
  • 70 Downloads / 205 Views

DOWNLOAD

REPORT


ORIGINAL PAPER

Quality assessment of pan-sharpening methods in high-resolution satellite images using radiometric and geometric index Mahdi Hasanlou 1 & Mohammad Reza Saradjian 1

Received: 2 October 2013 / Accepted: 7 October 2015 # Saudi Society for Geosciences 2015

Abstract This paper focuses on quality assessment of fusion of multispectral (MS) images with high-resolution panchromatic (Pan) images. Since most existing quality assessments take the entire image into account simultaneously and generate some uncertainties, a novel and rather objective quality index has been proposed for image fusion. The index is comprised of geometric and radiometric parts. Both geometric and radiometric measurements are calculated using morphological algorithm applied on an edge image to create a mask which is used to separate highfrequency regions from low-frequency ones. The accuracy assessment is made using common existing criteria on geometric and radiometric segments, and then a weighted sum is calculated to generate radiometric and geometric index (RG index). Several commonly used fusion algorithms such as IHS, modified IHS, PCA, Gram-Schmidt, Brovey Transform, Ehlers, High-Pass Modulation, Schowengerdt and UNB were applied on a very high-resolution GeoEye-1 and WorldView-2 images. In order to perform quality assessment, methods of Spectral Angle Mapper, Structural SIMilarity, correlation coefficients and universal quality index for which the normalization were possible (for comparison purposes) were used. The utilized RG index showed that by separating spectral and spatial component quality measurement, the quality assessment is made on fused

* Mahdi Hasanlou [email protected] Mohammad Reza Saradjian [email protected] 1

School of Surveying and Geospatial Engineering, College of Engineering, University of Tehran, Postal Code: 1439957131, P. O. Box: 11155-4563, Tehran, Iran

images in a more distinct, explicit, accurate and objective manner. Keywords Image fusion . Multispectral imagery . Quality assessment . Geometric distortion . Radiometric quality . Geometric quality

Introduction The widespread and diverse range of image fusion methods require accurate quality assessment criteria for comparison of results obtained from different algorithms. Since human perception of fused image is of fundamental importance, subjective criteria have been widely used to evaluate performance of different image fusion methods (DadrasJavan and Samadzadegan 2014; Toet and Franken 2003). However, as the digital applications of fused images are being increased, objective assessment criteria are getting more attention. Objective performance assessment is a rather complicated issue due to the variety of different application requirements and the lack of a clearly defined ground-truth. Image fusion methods have often been evaluated by comparing ideal fused image to a reference image (Ghosh and Joshi 2013; Wang et al. 2005). The assessment of quality normally involves computation of a number of different image fusion quality indices such as cor