- Project ID - 13
Project Title - Fast Bilateral-Space Stereo for Synthetic Defocus - Github link - https://github.com/Digital-Image-Processing-IIITH/project-pix-it
- Team Members
- Apoorva Thirupathi (2019121012)
- Balavarun Pedapudi (2020101012)
- Jhanvi Shingala (2020701008)
- Mohd Omama (2020701006)
-
Main Goals - Create a depth map from stereo pair and use that depth to render a synthetically defocused image. Implementing the paper.
-
Synthetical defocussing without hardware. Converting stereo images to bilateral space and solve depth as an optimization problem. This depth map can be used to render a defocussing effect on the image.
The project will be developed using python, opencv. -
Result of the project would be to create a synthetically defocussed image from the stereo image pair. This algorithm should be efficient enough to be used without much latency.
- Milestones for the project and tentative deadlines for the same are:
- Creating simplified bilateral grid (24 Oct)
- Stereo matching (28 Oct)
- Loss function formulation (31 Oct)
- Bilateral solver (optimization in bilateral space) (7 Nov)
- Obtaining the defocussing image from the stereo images using the depth map.(10 Nov)
- Optimizing and testing the implementation.(18 Nov)
- The stereo images for the implementation can be obtained from: