MARVIN : high speed 3D imaging for seedling classification
The next generation of automated sorting machines for seedlings demands 3D models of the plants to be made at high speed and with high accuracy. In our system the 3D plant model is created based on the information of 24 RGB cameras. Our contribution is an image acquisition technique based on volumetric intersection which is capable of the required order of speed and accuracy and which can easily be calibrated by non-expert operators. The use of 24 cameras leads to a non-trivial calibration procedure. A calibration procedure has been developed which is based on commercial single-calibration routines, offering robustness. This will be incorporated into the MARVIN-machine in such a way that a non-expert operator can easily calibrate the system with minimal intervention. In this paper we show a proof of principle for the fast and accurate image acquisition method.
| Main Authors: | , , , , , |
|---|---|
| Format: | Article in monograph or in proceedings biblioteca |
| Language: | English |
| Published: |
Wageningen Academic Publishers
|
| Subjects: | Calibration, Image acquisition, Multi-camera, Seedling inspection, |
| Online Access: | https://research.wur.nl/en/publications/marvin-high-speed-3d-imaging-for-seedling-classification |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The next generation of automated sorting machines for seedlings demands 3D models of the plants to be made at high speed and with high accuracy. In our system the 3D plant model is created based on the information of 24 RGB cameras. Our contribution is an image acquisition technique based on volumetric intersection which is capable of the required order of speed and accuracy and which can easily be calibrated by non-expert operators. The use of 24 cameras leads to a non-trivial calibration procedure. A calibration procedure has been developed which is based on commercial single-calibration routines, offering robustness. This will be incorporated into the MARVIN-machine in such a way that a non-expert operator can easily calibrate the system with minimal intervention. In this paper we show a proof of principle for the fast and accurate image acquisition method. |
|---|