Single Image Reconstruction Based on Sparse Representation

Original implementation code by Jianchao Yang for single image super-resolution using sparse representation. The algorithm works by first partitioning high/low-resolution training images into patches, then training patch pairs into coupled dictionaries. Testing images are mapped to the low-resolution dictionary to obtain sparse coefficients, which are then multiplied with the high-resolution dictionary to reconstruct the final image. This implementation provides excellent reference material for students studying super-resolution algorithms and demonstrates practical dictionary learning applications.