CPC G06T 7/593 (2017.01) [H04N 2013/0081 (2013.01)] | 19 Claims |
1. A method of extracting depth information from a stereoscopic image pair, the method comprising implementing, by an image processing system, the following:
receiving at the image processing system the stereoscopic image pair, formed of a target image and an associated reference image; and
for each pixel of the target image, computing a final disparity cost vector having cost components corresponding to different disparities, storing the final disparity cost vector in association with that pixel, and assigning to that pixel a disparity corresponding to a lowest cost component of the final disparity cost vector, wherein the extracted depth information comprises the disparities assigned to the pixels of the target image;
wherein, for at least a subset of the pixels of the target image, the final disparity cost vector is computed for each of those pixels by computing, with respect to the reference image, a set of matching costs for that pixel and the different disparities, and combining the matching costs with the one or more final disparity cost vectors stored in association with one or more adjacent pixels of the target image,
wherein each cost component L(p, d) of a final disparity cost vector for pixel p is computed as:
![]() in which p−x means the pixel adjacent pixel p in the target image for direction x∈X, L(p−x, d) means the cost component for disparity d of the final disparity cost vector for pixel p−x, V(d, d′) is a penalty term that selectively penalizes changes in disparity, D is a set of possible disparities and X is a set of one or more direction vectors.
|