Fusion Processing and Quality Evaluation of Remote Sensing Images Based on the Integration of Different Transform Methods with IHS

Abstract

With the increasingly universal use of remote sensing data, related development is required to extract and enhance image information. In this paper, a SPOT image including multispectral and panchromatic bands and a Landsat5/TM image have been used to study fusion performance based on the integration of intensity, hue and saturation (IHS) and five transform methods: linear stretching, histogram matching, double histogram matching, wavelet transforming and integration filtering. The results show that the fusion images were enhanced in quality by combining information from two images, producing increased definition, and preserving the spectral fidelity of the input dataset using an appropriate fusion method. The difference in fusion performance among integration methods was obvious. As a whole, the integration of a wavelet transform and IHS had the comparative advantage because it could get a higher quality fusion image than with information capacity, definition or spectral fidelity of the input dataset. In addition, the results showed the simple integration of linear stretching and IHS transforms could produce better fusion images, and it is practicable for use by non-professionals for remote sensing in many software applications including ENVI, Matlab and Erdas. Meanwhile it could be simpler and more practicable for most users, especially those not adept with GIS technology.