Input: LR image , LR image patches’ size and HR image patches’ size , the degradation matrix . |
Output: HR image |
Step 1. Extract patches from LR image , follow the raster-scan order, and start from the upper-left corner |
(some pixel overlap in each direction is allowed). |
Step 2. Recover HR image patches iteratively by Steps 2.1 and 2.2, until the maximum iteration times |
or convergence is reached. |
Step 2.1 Self-similarity regulation step: |
Step 2.1.1. Use bicubic method to up scale the unrecovered LR patch to the same size as HR patch, defined as . |
Step 2.1.2. Searching for a similar sized patch in ’s neighbor: |
Step 2.1.2.1. Compute each searching patch’s SSE as the self-similarity prior , |
|
Step 2.1.2.2. Find the least SSE patch, and compare its SSE with the adaptive threshold |
. If , define this least SSE patch as the similar patch . |
Step 2.1.3. Use degradation matrix to down sample similar patch , define as . |
Step 2.1.4. Subtract from LR patch , and get the residual . |
Step 2.1.5. Recover the residual to using IRLS algorithm according (9). |
Step 2.1.6. Add the to , according to (10). |
Step 2.2 Sparse dictionary regulation step: update according to (11). |
Step 3. Ensemble all to recover HR image (if there is pixel overlap, the weighted average method is needed). |