Image analysis is only one of the step in the global data analysis pipeline:
- Image acquisition
- Image analysis
- Data analysis
Sometimes it is easier to improve previous or future steps rather than spend too much time on the image analysis in itself…
Improve image acquisition
-
Better lighting conditions
-
Keep same disctance between camera and object
-
Increase contrast between background and object
-
Reduce the number of artefact in the object
-
Try different sensors (visible, infrared, multispectral)
 |
A. Camera </br> B. Fixed distance </br> C. Black background </br> D. Scale
Improve image analysis
-
Calibrate your pipeline
- Find a robust ground-truth dataset
- manual measurements
- simulated images
- Estimate the error
Errors can be non linear and likely to increase with the conplexity of the images
Improve data processing
-
Filter out noisy data
-
Correctly identify outlyers