Video-Level Binocular Single Vision Based on Temporal Coherency Post Processing Open Access
To deal with the problem that low-dynamic range (LDR) displays cannot present enough contrast and details, and high-dynamic range (HDR) displays are too expensive for general public use, the image-level binocular tone-mapping framework was proposed in 2012. When extending from this image-level system to a video-level system, however, a new problem-the temporal coherence-must be addressed. As each frame in a video is processed separately and no temporal coherency is taken into account, the results can be quite disturbing for rapidly-varying dynamics in a video. This project proposes a new framework to solve this problem, by integrating the existing image-level framework and the temporal coherency algorithm. The method begins by processing the video analysis operator is processed to calculate two characteristics, key video value and current frame value, of the input HDR video. Then each frame of the HDR video is tone mapped by the photographic tone reproduction operator with a user-defined exposure, and modified by a post-processing operator. The result can be output as the left LDR video of the LDR video pair. Next, the HDR video frames are tone-mapped by the same tone-mapping operator but with a different user-defined exposure, and modified by the post- processing operator to create the right LDR video frames. Each processed right LDR frame and its counterpart left LDR frame are tested by the binocular viewing comfort predictor (BVCP). If this LDR video frame pair cannot pass the BVCP, an iteration that increases or decreases the right exposure will be implemented until a qualified right-frame image can be generated. Finally, in all final right exposures, we find the one that is the nearest to the fixed left exposure, and begin a final iteration to regenerate the right LDR frames and combine them into a video. This video will then be output as the right LDR video of the LDR video pair. In experimental tests for the new proposed framework, the ability of preserving the object perception and the relative levels of brightness are tested. The experimental data show that this proposed new framework can effectively solve the temporal coherency problem and generate binocular LDR videos without disturbing effects.
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.