0_Image_acquisition.md

Image acquisition

Materials :

• RGB camera, 8 MP resolution or higher
• Container plants. Avoid using green-colored container.
  • Depending on user needs and python proficiency, it may be possible to measure objects other than container plants but that is not covered in the documentation.
• Target pages: Print first 2 pages of materials\targets.pdf. Do not use two copies of page 1! Attach to clipboard or other rigid backing. Measure the distances between the center points of the coded targets and compare to materials\scalebars.csv. Correct the .csv file if needed.
• Floor area minimum 1 m x 1 m, as level as possible.
  • There is no maximum area, however if the area is longer than 3 meters, consider using additional target pages.
• floor covering (black or other non-plant color) (optional)

Considerations:

• If lighting conditions are not stable, such as on a partly cloudy day, the measurement quality could be affected. Stable lighting will help to create a high-quality 3D point cloud.
• Likewise, the presence of wind will negatively affect the quality of the point cloud. The plants should be as still as possible. Avoid doing image acquisition during windy weather, and consider temporarily shutting off ventilation if necessary.

(a) Setup image acquisition area

1. The height of plant container is needed by EasyDCP, so be sure to measure and record the height of the container in meters. If the container is on any riser platform, record the height of the riser and add it to the container height. EasyDCP expects all plants to have the same container height.
2. If using floor covering, place on the desired area to be used for image acquisition. If not using floor covering, ensure floor area is free of any materials that may be similar to plant color (e.g. weeds)
3. Define measurement area by placing the two printed target pages at opposite corners. [see figure 1] Page 1 in bottom-left corner, page 2 in top-right corner. Ensure target pages are level (especially page 1!) and not upside-down.
   • Note: The two outermost targets (1 and 8) define the boundaries of the point cloud region. This is enabled by default but can be disabled, and helps to reduce processing time. If plant canopies extend beyond the boundary box defined by the two outermost targets, ensure crop_by_targets = False in params.ini during EasyDCP_Creation.
   • Additional target pages (targets.pdf page 3 and onward) may be used. This may improve point cloud quality if the image acquisition area is large. Extra pages can be placed anywhere and Metashape will automatically detect them. Only the locations of pages 1 and 2 are fixed at the outermost corners of the area. Consider using additional pages if the image acquisition area is longer or wider than 2 meters.

(b) Image acquisition

1. Arrange a group of container plants within the measurement area, by placing them in a single row, or 2 rows using triangular spacing [see figure 1]. Ensure at least 10cm gap between all plant canopies. Note: The number of plants per group is only limited by the size of the measurement area and the size of the plants.

   

   Figure 1

2. Photograph the plants from above, in one or more rows parallel to plant row. We recommend 10-15 images per each meter of plant row at 1 meter camera distance from target. See figure 2 for example: 4 rows of images (blue rectangles) were captured over a 1 m x 2 m space, average 12 images per meter per row.

   

   Figure 2

3. Repeat steps 4 and 5 for all remaining plants. We recommend to keep the number of plants per group constant if possible, and keep plant spacing uniform across groups. See figure 1a and 1b.

4. Organize photos into folders by group. The root folder can be located anywhere on any drive. One folder per one group of plants. Even if there is only one group, follow this folder structure. Create a \skip\ folder and copy \materials\scalebars.csv and \materials\orientation.ini to \skip\. One level of nested folders is also supported, which may be useful for separating images taken from different angles.

   Example: for a given population, assume 9 plants divided into 3 groups of 3 plants each:

population001
├───group001
│   ├── IMG_0001.JPG
│   ├── IMG_0002.JPG
│   └── ...
├───group002
│   ├── IMG_0001.JPG
│   ├── IMG_0002.JPG
│   └── ...
├───group003
│   └── angle01
│   	├── IMG_0001.JPG
│   	└── ...
│   └── angle02
│   	├── IMG_0001.JPG
│   	└── ...
└───skip
    ├── scalebars.csv
    ├── orientation.ini
    └── [other files/folders to ignore]

5. Proceed to EasyDCP_Creation.