CVPR 2010 tutorial Computational photography
Coded Computational Imaging
June 14, 2010 · Half-day morning session
Overview
Capture more by changing how cameras sample light.
Computational imaging combines optics, sensors, illumination, and reconstruction algorithms to capture information beyond a conventional photograph. This tutorial covers angular sampling for light fields and digital refocusing, coded exposure for motion deblurring, aperture modulation for depth estimation, and computational illumination.
01
Course slides
Introduction · PDF
- History of images and imaging
- Introduction to computational imaging
- Course details
Assorted Pixels · PDF
- Dimensions of imaging
- Multiplexing and multi-dimensional sampling
- Assorted pixel cameras: intensity, color, polarization, depth, focus, angle, and time
- HDR imaging and videography
Coding and Modulation in Cameras · PDF
- Photo manipulation versus coded imaging
- Point spread function invertibility and invariance
- Handling motion blur and defocus blur
- Strobing cameras: high-speed imaging of periodic signals using compressive sensing
- Coded sampling for temporal super-resolution
Light Fields and Applications · PDF
- Introduction to light fields and capture methods
- Shield fields and glare reduction
- Reinterpretable imaging: varying resolution after capture
- BiDi screen for 3D gesture interaction
- Bokode: camera-based interaction from a distance
Computational Illumination · PDF
- Geometry acquisition
- Acquiring material properties and appearance
- Acquiring volumetric media
- High-speed vision
- Novel displays
Future Directions · PDF
- Imaging beyond computer vision
- Smart optics, smart sensors, and smart lighting
- Challenges in image authentication, privacy, and social impact
02