A positioning method for the optimal seam in binocular visual image stitching
A binocular depth-guided seam strategy combining geometry, color, and depth features improves image stitching naturalness by 37%.
Published Mar 12, 20201 citationPaper ↗
Only vote on papers you've read. Sign in to vote.
Abstract
Image stitching technology is an important technology for generating large field of view images, and it is also a research hotspot in the field of digital image processing. Binocular imaging technology is closely related to image stitching technology and is widely used in robot vision, 3D reconstruction and other fields. However, in the existing image stitching methods, when the image with the distant view and the near object are coexisted, due to the parallax, the near object often has partial missing or unnatural transition, which leads to the poor image stitching effect and low naturalness. Aiming at this problem, we uses the characteristics of the binocular camera that can obtain depth information, and regards the information as one of the basis of the seam positioning. A strategy based on the combination of geometry, color and depth features is proposed, and we constructed a seaming algorithm model for positioning the optimal seam. The comparison experiment results show that the stitching naturalness is improved by 37% according to the proposed seaming algorithm.