This paper presents a non-iterative deep learning approach to compressive sensing (CS) image reconstruction using a convolutional autoencoder and a residual learning network. An efficient measurement design is proposed in order to enable training of the compressive sensing models on normalized and mean-centred measurements, along with a practical network initialization method based on principal component analysis (PCA). Finally, perceptual residual learning is proposed in order to obtain semantically informative image reconstructions along with high pixel-wise reconstruction accuracy at low measurement rates.
Pub. online:1 Jan 2003Type:Research ArticleOpen Access
Volume 14, Issue 4 (2003), pp. 431–444
In this paper, we shall propose a new method for the copyright protection of digital images. To embed the watermark, our new method partitions the original image into blocks and uses the PCA function to project these blocks onto a linear subspace. There is a watermark table, which is computed from projection points, kept in our new method. When extracting a watermark, our method projects the blocks of the modified image by using the PCA function. Both the newly projected points and the watermark table are used to reconstruct the watermark. In our experiments, we have tested our scheme to see how it works on original images modified by JPEG lossy compression, blurring, cropping, rotating and sharpening, and the experimental results show that our method is very robust and indeed workable.