A Generic Real Time Autoencoder-Based Lossy Image Compression

Abstract

Multimedia compression is a fundamental and significant research topic in the industrial field in the past several decades attempting to improve compression techniques. It is always a trade-off between size and quality where the growth rate of image, audio and video data is far beyond the improvement of the compression ratios achieved so far. Here, we are aiming to explore the potential of neural networks to achieve data compression, making use of multilayer neural networks providing a more efficient solution. In this paper, we present a lossy compression architecture, which utilizes the advantages of convolutional autoencoder (CAE) to replace the conventional transforms. Experimental results demonstrate that our method outperforms traditional coding algorithms, by achieving better compression ratios over the related work.