US 12,394,023 B2
Image denoising method and apparatus, electronic device, and storage medium
Jisong Ai, Guangdong (CN); Ke Xu, Guangdong (CN); Dehui Kong, Guangdong (CN); Ning Wang, Guangdong (CN); Xin Liu, Guangdong (CN); Jing You, Guangdong (CN); and Fang Zhu, Guangdong (CN)
Assigned to SANECHIPS TECHNOLOGY CO., LTD., Guangdong (CN)
Appl. No. 17/921,271
Filed by SANECHIPS TECHNOLOGY CO., LTD., Guangdong (CN)
PCT Filed Apr. 22, 2021, PCT No. PCT/CN2021/088927
§ 371(c)(1), (2) Date Oct. 25, 2022,
PCT Pub. No. WO2021/218765, PCT Pub. Date Nov. 4, 2021.
Claims priority of application No. 202010346311.9 (CN), filed on Apr. 27, 2020.
Prior Publication US 2023/0230206 A1, Jul. 20, 2023
Int. Cl. G06T 5/70 (2024.01); G06T 5/20 (2006.01)
CPC G06T 5/70 (2024.01) [G06T 5/20 (2013.01); G06T 2207/20081 (2013.01)] 11 Claims
OG exemplary drawing
 
1. An image denoising method, comprising:
acquiring an image to be processed; and
inputting the image to be processed into an image denoising model to acquire a denoised image,
wherein the image denoising model is a model formed by combining a U-shaped network, a residual network and a dense network,
wherein the image denoising model comprises: an input layer, a first convolutional layer, at least one dense residual module, a dense residual block, at least one upsampling module, a second convolutional layer, a third convolutional layer, and an output layer which are connected in sequence;
wherein a subtraction operation is performed on an output of the input layer and an output of the third convolutional layer, and a result of the subtraction operation is input to an input of the output layer;
an addition operation is performed on an output of the first convolutional layer and an output of the second convolutional layer, and a result of the addition operation is input to an input of the third convolutional layer;
the at least one dense residual module comprises a first dense residual submodule and a convolution submodule which are connected in sequence;
the at least one upsampling module comprises an upsampling submodule and a second dense residual submodule which are connected in sequence; and
an addition operation is performed on an output of the first dense residual submodule and an input of the upsampling submodule.