| 研究実績の概要 |
Summary of Research achievement. I have finished the dataset preparation, and on the basis of the dataset, I developped the multi-temporal SAR-optical method. The propose method is utilized to predict the post-disaster optical image. Our purpose is to predict the post-disaster optical image with landslide details, from the input of pre-disaster SAR-optical image pairs and post-disaster SAR image. Previous deep learning based methods can recover the optical image in good visual, unfortunately with the landslides disappeared. To reconstruct physically meaningful details, I calculate a weight matrix of post-disaster SAR image to measure the importance of each pixel, and utilize the weight matrix to guild the reconstruction of optical image. On the basis of the related works, we also developped several image quality improvement methods for remote sensing image denoising, resotration, and reconstruction. The related publications include 1 paper accepted by CVPR2019, 1 paper accepted by IEEE Transactions on Cybernetics, 1 paper accpepted by IEEE transactions on image processing, 1 paper accpeted by ISPRS journal, and 3 papers accepted by IEEE transactions on geoscience and remote
sensing.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
For this project, I have finished the data preparation and GAN generation for the fist stage. The next step is to apply the related methods to the disaster response application. On the basis of this project, the publications include 1 paper accepted by CVPR2019, 1 paper accepted by IEEE Transactions on Cybernetics (TCYB), 1 paper accpepted by IEEE transactions on image processing, and 3 papers accepted by IEEE transactions on geoscience and remote sensing.
|
| 今後の研究の推進方策 |
1. We plan to Apply the recovered SAR and optical satellite images to the disaster damage mapping. We will adopt the proposed SAR2Optical model to recover the pre- and post- disaster remote sensing image pairs, which will be adopted for the disaster damage mapping, including debris flow mapping, post-earthquake damage building mapping and flood area estimation.
2. Promote more general image translation methods, by considering self-supervised learning and attention learning.
3. Target at more challenging journals/conferences such IEEE Transactions on Pattern Analysis and Machine Intelligence and CVPR.
|
