Constructing High Resolution Satellite Image Using Interpolation Algorithm And Fast Nlm Filter

Abstract: In remote sensing, resolution of an image is limited by diffraction, atmospheric distortion and imperfect focus. Our proposed Resolution Enhancement(RE) scheme uses Dual tree-complex wavelet transform(DT-CWT), Lanczos interpolation and Fast NLM filtering for post processing. Fast Non Local Means filtering speed up the processing time compared to the NLM filtering. The Lanczos interpolator is used to interpolate the high frequency sub- bands and low resolution input image. Now these high frequency sub-bands are passed through Fast NLM filter for good clarity. Inverse DT-CWT is applied to obtain resolution enhanced image. In parallel performance of Fast NLM filter is also calculated for different set of input images using Mean Square Error(MSE) and Signal-to-noise ratio(SNR).

Keywords: Lanczos interpolator, Fast NLM filter, Mean Square Error(MSE) and Signal-to-noise ratio(SNR), Summed Square Index (SSI),Dual tree-complex wavelet transform(DT-CWT).