ADMIRE: a locally adaptive single-image, non-uniformity correction and denoising algorithm: application to uncooled IR camera
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929579150540800 |
|---|---|
| author | Tendero, Yohann Gilles, Jerome |
| author_facet | Tendero, Yohann Gilles, Jerome |
| contents | We propose a new way to correct for the non-uniformity (NU) and the noise in uncooled infrared-type images. This method works on static images, needs no registration, no camera motion and no model for the non uniformity. The proposed method uses an hybrid scheme including an automatic locally-adaptive contrast adjustment and a state-of-the-art image denoising method. It permits to correct for a fully non-linear NU and the noise efficiently using only one image. We compared it with total variation on real raw and simulated NU infrared images. The strength of this approach lies in its simplicity, low computational cost. It needs no test-pattern or calibration and produces no "ghost-artefact". |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_03615 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | ADMIRE: a locally adaptive single-image, non-uniformity correction and denoising algorithm: application to uncooled IR camera Tendero, Yohann Gilles, Jerome Image and Video Processing Computer Vision and Pattern Recognition We propose a new way to correct for the non-uniformity (NU) and the noise in uncooled infrared-type images. This method works on static images, needs no registration, no camera motion and no model for the non uniformity. The proposed method uses an hybrid scheme including an automatic locally-adaptive contrast adjustment and a state-of-the-art image denoising method. It permits to correct for a fully non-linear NU and the noise efficiently using only one image. We compared it with total variation on real raw and simulated NU infrared images. The strength of this approach lies in its simplicity, low computational cost. It needs no test-pattern or calibration and produces no "ghost-artefact". |
| title | ADMIRE: a locally adaptive single-image, non-uniformity correction and denoising algorithm: application to uncooled IR camera |
| topic | Image and Video Processing Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2411.03615 |