We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

eess.IV

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Image and Video Processing

Title: Image Compression with Recurrent Neural Network and Generalized Divisive Normalization

Abstract: Image compression is a method to remove spatial redundancy between adjacent pixels and reconstruct a high-quality image. In the past few years, deep learning has gained huge attention from the research community and produced promising image reconstruction results. Therefore, recent methods focused on developing deeper and more complex networks, which significantly increased network complexity. In this paper, two effective novel blocks are developed: analysis and synthesis block that employs the convolution layer and Generalized Divisive Normalization (GDN) in the variable-rate encoder and decoder side. Our network utilizes a pixel RNN approach for quantization. Furthermore, to improve the whole network, we encode a residual image using LSTM cells to reduce unnecessary information. Experimental results demonstrated that the proposed variable-rate framework with novel blocks outperforms existing methods and standard image codecs, such as George's ~\cite{002} and JPEG in terms of image similarity. The project page along with code and models are available at this https URL
Comments: Accpeted at IEEE CVPR Workshop
Subjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
Report number: 10.1109/CVPRW53098.2021.00209
Cite as: arXiv:2109.01999 [eess.IV]
  (or arXiv:2109.01999v1 [eess.IV] for this version)

Submission history

From: Khawar Islam Mr [view email]
[v1] Sun, 5 Sep 2021 05:31:55 GMT (439kb,D)

Link back to: arXiv, form interface, contact.