#
- Recording scenes as multiple sequential images with epsilon variations of the camera parameters
- Reconstruct final images from these observations
- Final image may look distorted to a human observer but it allows to decode valuable information about a scene
- Coded exposure: for preserving fast-moving objects info
- Coded aperture optical heterodyning: to record 4D of light and achieve digital refocusing
- Coded illumination: to recover depth info by using multiple flashes with a silhouette
- Coded sensitivity: to record changes in intensity of pixels to produce HDR
- The plenoptic function defines the ray of light in 3 spatial coordinates and 2 angles
- Using 2 points on 2 plane (r, s), (u, v) can represent any flow of light through an empty region of the space
- Since radiance is constant the photic field is 4 dimensional
- 4D reflectance field - the reflectance distribution in an opaque surface
- Making scenes more machine readable
- Overcoming limitations of digital photography like depth, resolution...
- Enables photos captured with more than one camera in more than one moment or angle
- Generalized optics: 4D ray bender that modifies a light field, less-less modification possible
- Generalized sensors: go beyond 2D capture of the scene and capture a 3D ray representation
- Generalized reconstruction: converstion of raw output into pixels, removing noise, dead pixelsxels, etc...
- Computational illumination: capture what you want to reveal in a scene programmatically
- Record scene via multiple images with epsilon variation on camera parameters (aperture, exposition, focus, movement...)
- Merged pic retains the best aspects of each, field of view , dynamic range, depth of field, spatial resolution...
- Avoid motion blur caused by moving cameras or objects
- Flutter the shutter open and closed during a coded exposure time
- The flutter changes the traditional box filter into a broadband filter to decode details of the blurred image
- We use a pattern alternated mask to encode the light field entering the camera
- Modulation of the light field possible
- Focus at different depths via computational techniques
- Capture multiple images with different exposures and merge them to obtain higher fidelity image
- Also feasible with sensors of mulitple (codeable) sensitivies for each cell