In order to solve the problem of reducing the storage space and image resolution after image compression, a method of compression and reconstruction of encrypted digital image based on deep super resolution ( SR) model was proposed. Firstly, the encrypted digital image was segmented, and the image sub-blocks after segmentation were encoded and compressed. Secondly, the classical SR reconstruction method ( sparse coding method) was combined with deep learning ( convolutional neural network) to construct a deep SR model, and this model was used to compress and decompress the image. Finally, the digital image after decryption was reconstructed. The results show that the image occupies less storage space after compression than before compression, and the compression effect is improved. The resolution of digital image after deep SR model reconstruction is relatively higher, and the peak signal-to-noise ratio is higher.
关键词
深度SR模型/加密数字图像/压缩/重构
Key words
deep super-resolution model/encrypt digital images/compression/reconfiguration
维普
万方数据