Viewing entries in
Computer Vision
IMAGE BASED MODELING ON A CONSTRAINED AXIS Abstract: Given a set of multiple calibrated photographs, silhouettes can be extracted to form a coarse approximation of 3D geometry. This image based model can further be refined by using feature based photoconsistency constraints and iterative bundle adjustment to correlate and refine the details found in the data set. Image based modeling has practical applications in visual effects, medical imaging, and prototyping.
Download full paper (pdf)
A COMPARISON OF TECHNIQUES FOR ESTIMATING DEPTH USING STEREO PAIRS Abstract: Disparity maps can provide a visual estimate of depth information generated by a pair of stereo images. The goal when producing disparity maps is to achieve a smooth and detailed map while minimizing error and maximizing efficiency. Three methods of varying complexity are explored to demonstrate and compare their results and efficiency on the Middlebury stereo test set [5].
Download full paper (pdf)
