eduzhai > Applied Sciences > Engineering >

Exploring Multi-Scale Feature Propagation and Communication for Image Super Resolution

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 23 pages

Abstract: Multi-scale techniques have achieved great success in a wide range ofcomputer vision tasks. However, while this technique is incorporated inexisting works, there still lacks a comprehensive investigation on variants ofmulti-scale convolution in image super resolution. In this work, we present aunified formulation over widely-used multi-scale structures. With thisframework, we systematically explore the two factors of multi-scale convolution-- feature propagation and cross-scale communication. Based on theinvestigation, we propose a generic and efficient multi-scale convolution unit-- Multi-Scale cross-Scale Share-weights convolution (MS$^3$-Conv). Extensiveexperiments demonstrate that the proposed MS$^3$-Conv can achieve better SRperformance than the standard convolution with less parameters andcomputational cost. Beyond quantitative analysis, we comprehensively study thevisual quality, which shows that MS$^3$-Conv behave better to recoverhigh-frequency details.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×