A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotyping
Abstract Background Automated segmentation of large amount of image data is one of the major bottlenecks in high-throughput plant phenotyping. Dynamic optical appearance of developing plants, inhomogeneous scene illumination, shadows and reflections in plant and background regions complicate automat...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2020-07-01
|
| Series: | Plant Methods |
| Subjects: | |
| Online Access: | http://link.springer.com/article/10.1186/s13007-020-00637-x |
| _version_ | 1818417317778817024 |
|---|---|
| author | Michael Henke Astrid Junker Kerstin Neumann Thomas Altmann Evgeny Gladilin |
| author_facet | Michael Henke Astrid Junker Kerstin Neumann Thomas Altmann Evgeny Gladilin |
| author_sort | Michael Henke |
| collection | DOAJ |
| description | Abstract Background Automated segmentation of large amount of image data is one of the major bottlenecks in high-throughput plant phenotyping. Dynamic optical appearance of developing plants, inhomogeneous scene illumination, shadows and reflections in plant and background regions complicate automated segmentation of unimodal plant images. To overcome the problem of ambiguous color information in unimodal data, images of different modalities can be combined to a virtual multispectral cube. However, due to motion artefacts caused by the relocation of plants between photochambers the alignment of multimodal images is often compromised by blurring artifacts. Results Here, we present an approach to automated segmentation of greenhouse plant images which is based on co-registration of fluorescence (FLU) and of visible light (VIS) camera images followed by subsequent separation of plant and marginal background regions using different species- and camera view-tailored classification models. Our experimental results including a direct comparison with manually segmented ground truth data show that images of different plant types acquired at different developmental stages from different camera views can be automatically segmented with the average accuracy of $$93\%$$ 93 % ( $$SD=5\%$$ S D = 5 % ) using our two-step registration-classification approach. Conclusion Automated segmentation of arbitrary greenhouse images exhibiting highly variable optical plant and background appearance represents a challenging task to data classification techniques that rely on detection of invariances. To overcome the limitation of unimodal image analysis, a two-step registration-classification approach to combined analysis of fluorescent and visible light images was developed. Our experimental results show that this algorithmic approach enables accurate segmentation of different FLU/VIS plant images suitable for application in fully automated high-throughput manner. |
| first_indexed | 2024-12-14T12:04:52Z |
| format | Article |
| id | doaj.art-9c2254cc8f5d4b61a2d22e623f4d1727 |
| institution | Directory Open Access Journal |
| issn | 1746-4811 |
| language | English |
| last_indexed | 2024-12-14T12:04:52Z |
| publishDate | 2020-07-01 |
| publisher | BMC |
| record_format | Article |
| series | Plant Methods |
| spelling | doaj.art-9c2254cc8f5d4b61a2d22e623f4d17272022-12-21T23:01:54ZengBMCPlant Methods1746-48112020-07-0116111010.1186/s13007-020-00637-xA two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotypingMichael Henke0Astrid Junker1Kerstin Neumann2Thomas Altmann3Evgeny Gladilin4Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)Abstract Background Automated segmentation of large amount of image data is one of the major bottlenecks in high-throughput plant phenotyping. Dynamic optical appearance of developing plants, inhomogeneous scene illumination, shadows and reflections in plant and background regions complicate automated segmentation of unimodal plant images. To overcome the problem of ambiguous color information in unimodal data, images of different modalities can be combined to a virtual multispectral cube. However, due to motion artefacts caused by the relocation of plants between photochambers the alignment of multimodal images is often compromised by blurring artifacts. Results Here, we present an approach to automated segmentation of greenhouse plant images which is based on co-registration of fluorescence (FLU) and of visible light (VIS) camera images followed by subsequent separation of plant and marginal background regions using different species- and camera view-tailored classification models. Our experimental results including a direct comparison with manually segmented ground truth data show that images of different plant types acquired at different developmental stages from different camera views can be automatically segmented with the average accuracy of $$93\%$$ 93 % ( $$SD=5\%$$ S D = 5 % ) using our two-step registration-classification approach. Conclusion Automated segmentation of arbitrary greenhouse images exhibiting highly variable optical plant and background appearance represents a challenging task to data classification techniques that rely on detection of invariances. To overcome the limitation of unimodal image analysis, a two-step registration-classification approach to combined analysis of fluorescent and visible light images was developed. Our experimental results show that this algorithmic approach enables accurate segmentation of different FLU/VIS plant images suitable for application in fully automated high-throughput manner.http://link.springer.com/article/10.1186/s13007-020-00637-xGreenhouse plant phenotypingVisible light imagingFluorescence imagingMultimodal image alignmentSupervised image segmentationMachine learning |
| spellingShingle | Michael Henke Astrid Junker Kerstin Neumann Thomas Altmann Evgeny Gladilin A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotyping Plant Methods Greenhouse plant phenotyping Visible light imaging Fluorescence imaging Multimodal image alignment Supervised image segmentation Machine learning |
| title | A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotyping |
| title_full | A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotyping |
| title_fullStr | A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotyping |
| title_full_unstemmed | A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotyping |
| title_short | A two-step registration-classification approach to automated segmentation of multimodal images for high-throughput greenhouse plant phenotyping |
| title_sort | two step registration classification approach to automated segmentation of multimodal images for high throughput greenhouse plant phenotyping |
| topic | Greenhouse plant phenotyping Visible light imaging Fluorescence imaging Multimodal image alignment Supervised image segmentation Machine learning |
| url | http://link.springer.com/article/10.1186/s13007-020-00637-x |
| work_keys_str_mv | AT michaelhenke atwostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT astridjunker atwostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT kerstinneumann atwostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT thomasaltmann atwostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT evgenygladilin atwostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT michaelhenke twostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT astridjunker twostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT kerstinneumann twostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT thomasaltmann twostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping AT evgenygladilin twostepregistrationclassificationapproachtoautomatedsegmentationofmultimodalimagesforhighthroughputgreenhouseplantphenotyping |