Cloud-Based Image Compression via Subband-Based Reconstruction

In this paper, we propose a novel cloud-based image compression scheme. In contrast to traditional compression schemes, our scheme targets a new scenario that ensures large scale images in the cloud are always available. Thus, our proposed scheme makes use of not only the internal correlation, but also the external correlation between the target image and images in the cloud to achieve advanced coding performance. One the encoder side, two kinds of information, global and in-band local information, are extracted from the wavelet pyramids of the input image and compressed accordingly. On the decoder side, a subband-based reconstruction approach is proposed for hierarchically generating the reconstructed image from the compressed information and a large dataset in the cloud. Experimental results demonstrate that our proposed scheme achieves high objective and subjective qualities at high compression ratios given a large scale image dataset.