Compression perception reconstruction method and system based on depth residual error network

Abstract

The invention relates to a compression perception reconstruction method and system based on a depth residual error network; the method comprises the following steps: obtaining original image signals as training data, and using scale transformation and segmentation treatment so as to segment the training data into a plurality of image blocks; obtaining a measuring value corresponding to the brightness component according to the brightness component of each image block and a compression perception theory model; using a fully connected network to make linear mapping treatment for the measuring value, thus obtaining a primary reconstruction result; inputting the primary reconstruction result into the depth residual error network, and training to obtain an estimated residual error value; fusingthe estimated residual error value with the primary reconstruction result so as to form reconstruction signals. The method introduces the depth residual error network so as to participate the signalreconstructions, thus restoring and reconstructing from the measuring value to the image, and using the difference between the learning and the target of the depth residual error network to improve the quality of the restored signals.

Description

technical field [0001] The present invention relates to the technical field of digital imaging, in particular to a compression sensing reconstruction method and system based on a deep residual network. Background technique [0002] In traditional digital imaging technology, according to Shannon’s sampling theorem (if you want to restore the original analog signal from the collected digital signal, the sampling frequency of the signal must be greater than or equal to twice the highest frequency of the signal), it is necessary to first analyze the scene or signal Fully sampled and then compressed to facilitate signal transmission and storage, such as figure 1 As shown, such a signal processing method of sampling first and then compressing brings a lot of waste in sampling and calculation. [0003] As a brand-new sampling theory, compressed sensing theory was proposed by Candes et al. in 2006. The core idea is to use the measurement matrix to perform a small number of non-adap...

AI Technical Summary

Problems solved by technology

[0013] Aiming at the problems of low reconstruction efficiency and poor reconstruction quality currently existing in the compressed sensing reconstruction technology, the present invention proposes a compressed sensing reconstruction method based on a deep residual network, which includes:

Images