Creating disparity and point clouds based on stereo endoscopy videos from robotic surgery
Evaluated on the Hamlyn Dataset
- Uses Stereo Block Matching algorithms to estimate disparity map
- Uses data from images passed through a Gaussian Adaptive Threshold to isolate blood vessels
- Cross matches blocks from left and right pairs alternating with blood vessel-thesholded images
- Produces a real time disparity map of the surgery site
Dependencies: OpenCV Contrib Numpy Sklearn
Examples of output:
