Ji Bian
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Ji Bian has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Fansheng Chen
Hangzhou Institute for Advanced Study, University of Chinese Academy of Sciences, Hangzhou 310024, China
2 shared publications - Qiyao Wang
Key Laboratory of Intelligent Infrared Perception, Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai, China
2 shared publications Lu Yin
Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai, China
1 shared publication- Min Zhu
Key Laboratory of Intelligent Infrared Perception, Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai, China
1 shared publication - Xue Zhao
Key Laboratory of Intelligent Infrared Perception, Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai, China
1 shared publication Yun Hou
Shanghai Institute of Technical Physics, Chinese Academy of Sciences, Shanghai, China
1 shared publication
Latest Publications
- 10 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 21 (1545598X) , 1-5
Detection and Long-Term Analysis of Anomalous Pixels Based on Scene and On-Orbit Radiometric Traceability
In the on-orbit images of the SDGSAT-1 TIS, there are image striping phenomena caused by the nonlinear-response of the detector and flickering effects caused by the detector’s jump response. Based on the characteristics of the SDGSAT-1 TIS, a detection method for anomalous pixels using time-domain gradients and nonlinearity is proposed. This method also includes adaptive threshold adjustment and on-orbit data detection based on scene images. The results show that in each band, the number of flickering pixels with a detection probability exceeding 95% is fewer than 7, and the relative positions of these flickering pixels in the images are fixed. Additionally, traceability analysis of the on-orbit radiometric data reveals that the noise deviation of flickering pixels is more than 10% greater compared to normal pixels. This indicates that the flickering behavior is not random and that these pixels are more prone to flickering than others. In each band, the number of nonlinear-response pixels with a detection probability exceeding 95% is fewer than 6. Additionally, a traceability analysis of on-orbit radiometric data shows that the response deviation of nonlinear-response pixels is more than 0.7% compared to normal pixels. This result indicates that the occurrence of nonlinear-response pixels on orbit is not random and provides a new approach for early detection of anomalous pixels based on on-orbit radiometric data.
- 10 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 21 (1545598X) , 1-5
Detection and Long-Term Analysis of Anomalous Pixels Based on Scene and On-Orbit Radiometric Traceability
In the on-orbit images of the SDGSAT-1 TIS, there are image striping phenomena caused by the nonlinear-response of the detector and flickering effects caused by the detector’s jump response. Based on the characteristics of the SDGSAT-1 TIS, a detection method for anomalous pixels using time-domain gradients and nonlinearity is proposed. This method also includes adaptive threshold adjustment and on-orbit data detection based on scene images. The results show that in each band, the number of flickering pixels with a detection probability exceeding 95% is fewer than 7, and the relative positions of these flickering pixels in the images are fixed. Additionally, traceability analysis of the on-orbit radiometric data reveals that the noise deviation of flickering pixels is more than 10% greater compared to normal pixels. This indicates that the flickering behavior is not random and that these pixels are more prone to flickering than others. In each band, the number of nonlinear-response pixels with a detection probability exceeding 95% is fewer than 6. Additionally, a traceability analysis of on-orbit radiometric data shows that the response deviation of nonlinear-response pixels is more than 0.7% compared to normal pixels. This result indicates that the occurrence of nonlinear-response pixels on orbit is not random and provides a new approach for early detection of anomalous pixels based on on-orbit radiometric data.
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 21 (1545598X) , 1-5
Nonlinear Response Correction Based on Fully Connected Neural Network
Imagery from the Thermal Infrared Spectrometer (TIS) of the first Sustainable Development Goals Science Satellite (SDGSAT-1) provides a high-resolution observation view of ground objects, while the nonlinear response between imaging modules affects the quality of the image. Using the thermal infrared data from SDGSAT-1 TIS, an effective method of inter-module nonlinear response correction is presented in this paper. Firstly, an improved frequency domain guided LRSID algorithm is utilized to eliminate the fringe noise in the module, in conjunction with a boundary condition to avoid distortion at the boundary of the module. Next, a nonlinear response correction method utilizing a fully connected neural networks is developed to compensate for the difference between modules based on overlapping pixels. The correction parameters are calculated by the network trained on overlapping pixels from noiseless images. Results show that this method has produced considerable improvements in visual effects, and the distortion in response between modules was controlled below 0.5%, which is comparable with the work for Landsat 8. Besides, The Edge Slope (ES) of the corrected image makes almost no reduction, which means the nonlinear response is corrected without sacrificing image quality.
- 19 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Letters
- Vol. 21 (1545598X) , 1-5
Nonlinear Response Correction Based on Fully Connected Neural Network
Imagery from the Thermal Infrared Spectrometer (TIS) of the first Sustainable Development Goals Science Satellite (SDGSAT-1) provides a high-resolution observation view of ground objects, while the nonlinear response between imaging modules affects the quality of the image. Using the thermal infrared data from SDGSAT-1 TIS, an effective method of inter-module nonlinear response correction is presented in this paper. Firstly, an improved frequency domain guided LRSID algorithm is utilized to eliminate the fringe noise in the module, in conjunction with a boundary condition to avoid distortion at the boundary of the module. Next, a nonlinear response correction method utilizing a fully connected neural networks is developed to compensate for the difference between modules based on overlapping pixels. The correction parameters are calculated by the network trained on overlapping pixels from noiseless images. Results show that this method has produced considerable improvements in visual effects, and the distortion in response between modules was controlled below 0.5%, which is comparable with the work for Landsat 8. Besides, The Edge Slope (ES) of the corrected image makes almost no reduction, which means the nonlinear response is corrected without sacrificing image quality.