For my project, I have to handle image differences with OpenCV. The goal is to detect intrusion into the zone.
To be more clear, here are the ins and outs:
Inputs
- Link Image
- The second image is from about the same point of view (may be an error field)
Outputs:
- Detection of new objects in the scene.
Bonus:
- Recognition of these objects.
For me, the most difficult part of this is to remove small differences (luminosity, camera position field error, tree movement ...)
I already read a lot about OpenCV image processing (subtraction, erosion, threshold, SIFT, SURF ...) and has good results.
What I would like is a list of steps that you think are best for good detection (people, cars ...) and algorithms for each step.
Many thanks for your help.