A Joint Intensity and Depth Co-Sparse Analysis Model for Depth Map Super-Resolution
This work addresses depth map super-resolution for computer vision applications, but it is incremental as it builds on existing co-sparse analysis models by extending them to bimodal data.
The paper tackles the problem of inferring high-resolution depth maps from low-resolution depth measurements and a high-resolution intensity image by introducing a bimodal co-sparse analysis model that captures interdependencies between intensity and depth. It proposes a method to learn analysis operators from training data, leading to an efficient algorithm for depth map super-resolution.
High-resolution depth maps can be inferred from low-resolution depth measurements and an additional high-resolution intensity image of the same scene. To that end, we introduce a bimodal co-sparse analysis model, which is able to capture the interdependency of registered intensity and depth information. This model is based on the assumption that the co-supports of corresponding bimodal image structures are aligned when computed by a suitable pair of analysis operators. No analytic form of such operators exist and we propose a method for learning them from a set of registered training signals. This learning process is done offline and returns a bimodal analysis operator that is universally applicable to natural scenes. We use this to exploit the bimodal co-sparse analysis model as a prior for solving inverse problems, which leads to an efficient algorithm for depth map super-resolution.