In this chart; we represent the chronological order for the flow of ideas in Face Hallucination (FH). This doesn’t necessarily mean that these are the only milestones along the road, but this reflects the flow we followed in our research. This was influenced by our college mentors advice and our limited experience in the area subject to research. On the other hand, Google search results showed some of these papers on the first page, which also influenced our path in a way.
In year 2000, the pioneers Baker and Kanade coined the term for face hallucination. Then a framework to guarantee a successful face hallucination was postulated by Ce Liu et al. Dual dictionary learning proposed by Yang et al. started a wave of contributions in the area of sparse modeling. Hu et al. made use of sparsity and based their work on the assumption that similar face images share a similar structure. Then, they used a K-Nearest-Neighbors search and a single overcomplete dictionary to reconstruct the target high resolution face. Using a similar assumption an enhanced approach was proposed later by Li et al. The same stream of ideas was also used for video surveillance by Jia et al.
If the presentation doesn’t start automatically, please click here [presentation].