<abstract xmlns="http://www.w3.org/1999/xhtml">

<p>In this paper, semantic segmentation based on a convolutional neural network is used to guide the image color restoration process. In order to avoid the problem of content mismatch, higher-order features are first extracted from the basic dimensions of the input image, and the convolutional operation is done on the feature map by the excitation function. Then the network parameters are optimized and updated by the backpropagation algorithm to minimize the error between the prediction result and the real result. Finally, in image processing, the dilated convolution technique is introduced to perform noise reduction on the semantic segmented image pixels to optimize the restoration quality. In order to prove the effectiveness of the method in this paper, an experimental analysis of the method is conducted. The experimental results show that the peak signal-to-noise ratio of the model proposed in this paper is higher than 42.986db on average, the structural similarity reaches 0.8%, and the fit is around 0.75. And its color difference can reach at least 11.2% with the increase of iterations. It indicates that the reduction algorithm of semantic segmentation and convolutional neural network greatly improves the accuracy of color reduction results and can obtain printmaking images with higher color quality.</p>
</abstract>

In this paper, semantic segmentation based on a convolutional neural network is used to guide the image color restoration process. In order to avoid the problem of content mismatch, higher-order features are first extracted from the basic dimensions of the input image, and the convolutional operation is done on the feature map by the excitation function. Then the network parameters are optimized and updated by the backpropagation algorithm to minimize the error between the prediction result and the real result. Finally, in image processing, the dilated convolution technique is introduced to perform noise reduction on the semantic segmented image pixels to optimize the restoration quality. In order to prove the effectiveness of the method in this paper, an experimental analysis of the method is conducted. The experimental results show that the peak signal-to-noise ratio of the model proposed in this paper is higher than 42.986db on average, the structural similarity reaches 0.8%, and the fit is around 0.75. And its color difference can reach at least 11.2% with the increase of iterations. It indicates that the reduction algorithm of semantic segmentation and convolutional neural network greatly improves the accuracy of color reduction results and can obtain printmaking images with higher color quality.

Research on virtual color restoration algorithm for printmaking art images based on semantic segmentation and convolutional neural network

Applied Mathematics and Nonlinear Sciences

This work is licensed under the Creative Commons Attribution 4.0 International License.

In this paper, semantic segmentation based on a convolutional neural network is used to guide the image color restoration process. In order to avoid the...

Research on virtual color restoration algorithm for printmaking art images based on semantic segmentation and convolutional neural network

Pubblicato online: 09 ott 2023

Pagine: -

Ricevuto: 16 nov 2022

Accettato: 19 apr 2023

DOI: https://doi.org/10.2478/amns.2023.2.00545

Parole chiave
Convolutional neural network, Semantic segmentation, Image virtual reduction, Excitation function, Backpropagation, Dilation convolution

© 2023 Yongbo Wang, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Research on virtual color restoration algorithm for printmaking art images based on semantic segmentation and convolutional neural network

Pubblicato online: 09 ott 2023

Pagine: -

Ricevuto: 16 nov 2022

Accettato: 19 apr 2023

DOI: https://doi.org/10.2478/amns.2023.2.00545

Parole chiaveConvolutional neural network, Semantic segmentation, Image virtual reduction, Excitation function, Backpropagation, Dilation convolution

© 2023 Yongbo Wang, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Parole chiave
Convolutional neural network, Semantic segmentation, Image virtual reduction, Excitation function, Backpropagation, Dilation convolution