Pansharpening Using Data Driven Model Based on Linear Regression

With the launching of many earth’s observation satellites, the amount of data capturing the Earth’s surface has been increasing to a great extent. In this paper, we emphasize the need for analyzing the satellite image data particularly in the context of data fusion applied to data taken from sensors of different resolution. The problem lies in maintaining the spectral characteristics of the multispectral images when panchromatic image is used to estimate the high spatial multispectral image. We take a wholesome approach based on the reflectance data irrespective of the sensor physics. The approach aims to produce an enhanced spatial resolution multispectral image having the same resolution as that of the panchromatic data while still preserving the spectral characteristics of the multispectral image. Using a linear regression model between multispectral and panchromatic data, an optimal solution in terms of Lagrange multiplier is provided and validated to maximize the spectral consistency of the fused image. The chi-square test is used to check the “goodness of fitd” of the data. The experimental results are discussed and presented using IKONOS satellite data.