Abstract
We proposed mask scale adjustment method for stereo matching inspired by volume perception in human visual perception. Stereo matching is required to obtain 3-dimensional (3D) information from stereo pair images, and many sophisticated approaches have been developed to precisely obtain correspondence in stereo pairs. We found a forgotten problem not solved by these methods inherent in stereo matching and originating in the scale difference between stereo pair images. We propose mask scale adjustment for improving matching performance, especially in adjacent areas of binocularly unpaired parts, on objects with curved surfaces. Implementing our proposal into conventional stereo matching would yield more precise 3D information on objects and realize of volume perception.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call