Super-resolution for remote sensing images using content adaptive detail enhanced self examples
S. Vishnukumar, M. Wilscy · 2016
This paper proposes a single image super-resolution (SR) technique for remote sensing images using content adaptive detail enhanced self examples. This method exploits large number of similar patches exists in the remote sensing images by using self examples. A high frequency layer is extracted from the input low resolution image and details of the high frequency layer are enhanced using content adaptive method to form the self examples. The root mean square difference of feature vectors extracted from self examples is given to a Gaussian function to find the weight. Weighted average of pixel computed using weights predicts the pixels of the high resolution high frequency layer. The reconstructed high resolution high frequency layer is combined with the linearly interpolated high resolution image to form the final high resolution image. Qualitative and Quantitative experimental analysis show that the proposed method gives better results than other existing methods. The results have better visual quality since image details are well preserved.