We study a transform coder that employs a dynamic programming based rate-distortion optimization framework for time segmentation. Although this coder exhibits a high performance, its computational complexity makes it unfeasible for many practical applications. It is investigated whether upfront time segmentation can reduce computational complexity without a significant increase in perceptual distortion. Upfront time segmentation can be accomplished by replacing the rate-distortion cost functional with low-complexity cost measures, that are independent of bit rate and perceptual distortion. Through both quantitative and qualitative evaluation it is shown that dynamic programming based upfront time segmentation for minimization of perceptual entropy can be a viable alternative to rate-distortion optimal time segmentation.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=13283
(592KB)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
