eduzhai > Applied Sciences > Engineering >

Early Exit or Not Resource-Efficient Blind Quality Enhancement for Compressed Images

  • Save

... pages left unread,continue reading

Document pages: 25 pages

Abstract: Lossy image compression is pervasively conducted to save communicationbandwidth, resulting in undesirable compression artifacts. Recently, extensiveapproaches have been proposed to reduce image compression artifacts at thedecoder side; however, they require a series of architecture-identical modelsto process images with different quality, which are inefficient andresource-consuming. Besides, it is common in practice that compressed imagesare with unknown quality and it is intractable for existing approaches toselect a suitable model for blind quality enhancement. In this paper, wepropose a resource-efficient blind quality enhancement (RBQE) approach forcompressed images. Specifically, our approach blindly and progressivelyenhances the quality of compressed images through a dynamic deep neural network(DNN), in which an early-exit strategy is embedded. Then, our approach canautomatically decide to terminate or continue enhancement according to theassessed quality of enhanced images. Consequently, slight artifacts can beremoved in a simpler and faster process, while the severe artifacts can befurther removed in a more elaborate process. Extensive experiments demonstratethat our RBQE approach achieves state-of-the-art performance in terms of bothblind quality enhancement and resource efficiency. The code is available atthis https URL.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×