Single image super resolution: An efficient approach using auto-learning and filter pooling
M. S. Greeshma, V. R. Bindu · 2017
Stimulated by recent approaches on internal database driven image super resolution, we propose a super resolution algorithm based on auto learning and filter pooling approach. This paper presents an algorithm where auto-trained high frequency cluster of patches are used to reconstruct a high resolution image. The auto-trained features are extracted using filter pooling approach combining the advantages of Sobel and Gabor filters. The key feature of this approach is preserving features at different frequencies with different orientations using self-similarity to avoid the use of multiple images. The experimental analysis shows that the proposed technique gives better results when compared to existing state-of-the-art internal database methods for image super resolution in terms of quantitative and qualitative performance measures.