Near lossless compressed image soft decoding method based on wide activation recurrent neural network

Abstract

The invention belongs to the technical field of digital image processing, and relates to an image soft decoding method, in particular to a near lossless compressed image soft decoding method based on a wide activation recurrent neural network, which comprises the following steps: S1, acquiring and preprocessing a sample; s2, constructing a wide activation recurrent neural network model; s3, training a wide activation recurrent neural network model; and S4, testing the wide activation recurrent neural network model. The problems that the quality of a compressed image reconstructed by an existing method is not clear enough, and the pixel boundary constraint is not strict enough are solved. According to the method, the near-lossless compressed image cyclic reconstruction model based on the wide activation neural network is constructed, and the pixel boundary constraint is redefined in a reasonable and efficient manner, so that the compressed image reconstruction quality is effectively improved, the strict pixel boundary constraint is ensured, the visual sense experience of people is improved, and the reconstruction efficiency of the compressed image is improved. The method can be used for completing soft decoding of near lossless compressed images.

Description

technical field [0001] The invention belongs to the technical field of digital image processing, and relates to an image soft decoding method, in particular to a near lossless compressed image soft decoding method based on a wide activation cycle neural network. Background technique [0002] In remote sensing imaging, telemedicine, Internet of Things and other fields, image encoders are usually packaged in highly integrated chips. Modifying a system-on-chip encoder to improve image quality is often a thorny problem. The advantage of image soft decoding is that it can improve the quality of compressed images without modifying the encoder. [0003] Soft decoding is actually a kind of ill-conditioned inverse problem. Previous soft decoding techniques are mostly based on explicit image modeling and optimization. Such as based on autoregressive model, based on sparse, random walk graph, etc. In recent years, with the popularity of convolutional neural networks, researchers ha...

AI Technical Summary

Problems solved by technology

[0008] The purpose of the present invention is to address the deficiencies in the above-mentioned prior art, to provide a near-lossless compressed image soft decoding method based on a wide activation cyclic neural network, to solve the problem that the quality of the reconstructed compressed image in the prior art is not clear enough, and the pixel boundary constraints not strict enough

Images