A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on Orbit
Signal-to-Noise Ratio (SNR) is the benchmark to evaluate the quality of optical remote sensors. For SNR estimation, most of the traditional methods have complicated processes, low efficiency, and general accuracy. In particular, they are not suitable for the distributed computation on intelligent sa...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-02-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/13/5/2851 |
| _version_ | 1797615799320444928 |
|---|---|
| author | Bo Zhu Xiaoning Lv Congao Tan Yuli Xia Junsuo Zhao |
| author_facet | Bo Zhu Xiaoning Lv Congao Tan Yuli Xia Junsuo Zhao |
| author_sort | Bo Zhu |
| collection | DOAJ |
| description | Signal-to-Noise Ratio (SNR) is the benchmark to evaluate the quality of optical remote sensors. For SNR estimation, most of the traditional methods have complicated processes, low efficiency, and general accuracy. In particular, they are not suitable for the distributed computation on intelligent satellites. Therefore, an intelligent SNR estimation algorithm with strong computing power and more accuracy is urgently needed. Considering the simplicity of distributed deployment and the lightweight goal, our first proposition is to design a convolutional neural network (CNN) similar to VGG (proposed by Visual Geometry Group) to estimate SNR for optical remote sensors. In addition, considering the advantages of multi-branch structures, the second proposition is to train the CNN in a novel method of multi-branch training and parameter-reconstructed inference. In this study, simulated and real remote sensing images with different ground features are utilized to validate the effectiveness of our model and the novel training method. The experimental results show that the novel training method enhances the fitting ability of the network, and the proposed CNN trained in this method has high accuracy and reliable SNR estimation, which achieves a 3.9% RMSE for noise-level-known simulated images. When compared to the accuracy of the reference methods, such as the traditional and typical SNR methods and the denoising convolutional neural network (DnCNN), the performance of the proposed CNN trained in a novel method is the best, which achieves a relative error of 5.5% for hyperspectral images. The study is fit for optical remote sensing images with complicated ground surfaces and different noise levels captured by different optical remote sensors. |
| first_indexed | 2024-03-11T07:32:00Z |
| format | Article |
| id | doaj.art-96768f18d4944045ba98b9d9a749e14c |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-11T07:32:00Z |
| publishDate | 2023-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-96768f18d4944045ba98b9d9a749e14c2023-11-17T07:15:55ZengMDPI AGApplied Sciences2076-34172023-02-01135285110.3390/app13052851A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on OrbitBo Zhu0Xiaoning Lv1Congao Tan2Yuli Xia3Junsuo Zhao4Institute of Software, Chinese Academy of Sciences, No. 4 Nan Si Street, Haidian District, Beijing 100089, ChinaInstitute of Software, Chinese Academy of Sciences, No. 4 Nan Si Street, Haidian District, Beijing 100089, ChinaSchool of Electrical and Information Engineering, Zhengzhou University, No. 100 Ke Xue Road, Zhengzhou 450001, ChinaInstitute of Software, Chinese Academy of Sciences, No. 4 Nan Si Street, Haidian District, Beijing 100089, ChinaInstitute of Software, Chinese Academy of Sciences, No. 4 Nan Si Street, Haidian District, Beijing 100089, ChinaSignal-to-Noise Ratio (SNR) is the benchmark to evaluate the quality of optical remote sensors. For SNR estimation, most of the traditional methods have complicated processes, low efficiency, and general accuracy. In particular, they are not suitable for the distributed computation on intelligent satellites. Therefore, an intelligent SNR estimation algorithm with strong computing power and more accuracy is urgently needed. Considering the simplicity of distributed deployment and the lightweight goal, our first proposition is to design a convolutional neural network (CNN) similar to VGG (proposed by Visual Geometry Group) to estimate SNR for optical remote sensors. In addition, considering the advantages of multi-branch structures, the second proposition is to train the CNN in a novel method of multi-branch training and parameter-reconstructed inference. In this study, simulated and real remote sensing images with different ground features are utilized to validate the effectiveness of our model and the novel training method. The experimental results show that the novel training method enhances the fitting ability of the network, and the proposed CNN trained in this method has high accuracy and reliable SNR estimation, which achieves a 3.9% RMSE for noise-level-known simulated images. When compared to the accuracy of the reference methods, such as the traditional and typical SNR methods and the denoising convolutional neural network (DnCNN), the performance of the proposed CNN trained in a novel method is the best, which achieves a relative error of 5.5% for hyperspectral images. The study is fit for optical remote sensing images with complicated ground surfaces and different noise levels captured by different optical remote sensors.https://www.mdpi.com/2076-3417/13/5/2851quality estimationsignal-to-noise ratioCNNmulti branch trainingparameter reconstruction |
| spellingShingle | Bo Zhu Xiaoning Lv Congao Tan Yuli Xia Junsuo Zhao A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on Orbit Applied Sciences quality estimation signal-to-noise ratio CNN multi branch training parameter reconstruction |
| title | A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on Orbit |
| title_full | A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on Orbit |
| title_fullStr | A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on Orbit |
| title_full_unstemmed | A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on Orbit |
| title_short | A Multi-Branch Training and Parameter-Reconstructed Neural Network for Assessment of Signal-to-Noise Ratio of Optical Remote Sensor on Orbit |
| title_sort | multi branch training and parameter reconstructed neural network for assessment of signal to noise ratio of optical remote sensor on orbit |
| topic | quality estimation signal-to-noise ratio CNN multi branch training parameter reconstruction |
| url | https://www.mdpi.com/2076-3417/13/5/2851 |
| work_keys_str_mv | AT bozhu amultibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT xiaoninglv amultibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT congaotan amultibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT yulixia amultibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT junsuozhao amultibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT bozhu multibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT xiaoninglv multibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT congaotan multibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT yulixia multibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit AT junsuozhao multibranchtrainingandparameterreconstructedneuralnetworkforassessmentofsignaltonoiseratioofopticalremotesensoronorbit |