Real Time Re-Calibration of Stereo Cameras

A method of re-calibrating a stereo camera comprising:

• receiving, by a processor, a pair of stereo image frames extracted from a stereoscopic video sequence, each frame included in the pair of stereo image frames divided into a plurality of image sections dimensioned in pixels;
• for each image section, transforming image data included in a first image section and image data included in a second image section into compressed image data, comparing compressed image data from the first image section to compressed image data from the second image section at multiple shift positions to obtain a disparity value in pixels at each shift position;
• for each disparity value, filtering the disparity value using a correlation filtering function determining a correlation coefficient for compressed image data included in the first image section and compressed image data included in the second image section, if the correlation coefficient passes a first filtering threshold, accepting the disparity value into a calibration profile, if the correlation coefficient fails the first filtering threshold, rejecting the disparity value from further analysis;
• and
• rendering, by a processor on a stereoscopic display device, the stereoscopic video sequence by adjusting a position of one or more image sections according to one or more disparity values included in a calibration profile and projecting contents of each image section corresponding to image pixels of the stereoscopic video sequence to display pixels of the stereoscopic display device.

A method of re-calibrating a stereoscopic 3D camera comprising:

• receiving, by a processor, a pair of stereoscopic video frames extracted from a stereoscopic video sequence, each frame in the pair of stereoscopic frames divided into a plurality of image sections dimensioned in pixels;
• for each image section, performing a disparity analysis comprising:
• comparing image data included in a first image section to image data included in a second image section at multiple shift positions to obtain a disparity value in pixels at each shift position;
• applying an aggregation function to the disparity value at each shift position to generate an aggregate disparity value;
• mapping the aggregate disparity value to a stereoscopic video frame by associating the aggregate disparity value with position data for the image section, the position data describing the location of the image section within the stereoscopic video frame;
• and
• adding the aggregate disparity value to a calibration profile;
• and
• rendering, by a processor on a stereoscopic display device, a stereoscopic video sequence by adjusting in real time one or more positions of one or more image sections based on one or more aggregate disparity values and projecting contents of the image sections corresponding to image pixels of the sequence of stereoscopic video frames to display pixels of the stereoscopic display device.