Feature Preserving Interpolation of Multispectral Images Using Local Binary Patterns

In this paper, we propose a novel technique for feature preserving interpolation of multi spectral (MS) images. Given an MS image and a panchromatic (PAN), high resolution MS image is obtained by estimating the details from the PAN image. We exploit the fact that the local geometry of the low resolution MS image is similar to that of the PAN image. The image features such as edges, corners, and curves carry crucial information in remote sensing applications. We models by using local binary patterns that best capture underlying geometric information and obtain high resolution enhancement for the MS images by learning discrete wavelet coefficients of the high resolution features present in PAN image. The proposed technique has been tested on various remotely sensed images. The results show the superiority of the proposed technique over existing interpolation techniques.