Biological image denoising method based on double-enhanced residual network

Abstract

The invention discloses a biological image denoising method based on a double-enhanced residual network, which comprises two sub-networks with the same structure, and each sub-network downscales and upscales features through a coding-decoding hierarchical structure, so that a GPU (Graphic Processing Unit) can generate a larger receptive field; in the encoding process, downsampling is carried out by adopting a convolutional layer to obtain image information, and a residual block is superposed to carry out preliminary extraction on features; in the decoding process, the upsampling feature capability of transposed convolution is utilized, residual dense blocks are superposed, deep features are extracted, and image details are recovered; in the two sub-networks, jump connection is added between a convolutional layer and a transposed convolutional layer, a supervision and attention module and four cross feature fusion modules are connected between the two sub-networks, and the connections are beneficial to transmission of image detail information, deepening of the networks and improvement of recovery performance at the same time.

Description

technical field [0001] The invention relates to an image denoising method, in particular to a biological image denoising method based on a double enhanced residual network. Background technique [0002] In the process of digital image acquisition and transmission, image degradation will be caused by noise. For biological data sets, noise may come from endogenous biological factors such as cell cycle and life history changes, or from exogenous technical factors such as sample preparation and instrument changes. Phenomena such as sharpness and fuzzy detailed features affect the accuracy of inferring the underlying process. [0003] Although traditional denoising methods can eliminate noise to a certain extent, it is usually difficult to preserve the details of the image in the restoration process. With the continuous development of deep learning technology in the field of image processing, many methods of using convolutional neural networks to improve image denoising perform...

AI Technical Summary

Problems solved by technology

[0002] In the process of digital image acquisition and transmission, image degradation will be caused by noise

For biological data sets, noise may come from endogenous biological factors such as cell cycle and life history changes, or from exogenous technical factors such as sample preparation and instrument changes. Phenomena such as clarity and fuzzy detailed features affect the accuracy of inferring the underlying process

[0003] Although traditional denoising methods can eliminate noise to a certain extent, it is usually difficult to preserve the details of the image during the restoration process.

With the continuous development of deep learning technology in the field of image processing, many methods of using convolutional neural networks to improve image denoising performance have been proposed, such as enriching the functions of neural networks by stacking the number of layers of the network. increase, the training model will degenerate

In addition, the depth features of the network and the output must maintain the same resolution, and the denoising network will be limited by the GPU memory in terms of the number of network layers and the number of parameters.

Images