Multi-Scale Deep Compressive Imaging

Recently, deep learning-based compressive imaging (DCI) has surpassed conventional compressive imaging in reconstruction quality and running speed. While multi-scale sampling has shown superior performance over single-scale, research in DCI has been limited to single-scale sampling. Despite training with single-scale images, DCI tends to favor low-frequency components similar to conventional multi-scale sampling, especially at low subrates. From this perspective, it would be easier for the network to learn multi-scale features with a multi-scale sampling architecture. In this work, we propose a multi-scale deep compressive imaging (MS-DCI) framework which jointly learns to decompose, sample, and reconstruct images at multi-scale. A three-phase end-to-end training scheme is introduced with an initial and two enhanced reconstruction phases to demonstrate the efficiency of multi-scale sampling and further improve the reconstruction performance. We analyze the decomposition methods (including pyramid, wavelet, and scale-space), sampling matrices, and measurements and show the empirical benefit of MS-DCI, which consistently outperforms both conventional and deep learning-based approaches.

Automated summary

Recent research has shown that deep learning-based multi-scale compressive imaging outperforms conventional methods in terms of reconstruction quality and speed, and the efficiency and performance of multi-scale sampling can be further improved by jointly learning to decompose, sample, and reconstruct images.