Early detection of anomalies in movement

Scope

In a population of objects, some can display anomalous movement patterns.
Most common cases are:

  • A/ The normal items move freely, the abnormal move little or not at all
  • B/ The normal items don’t move, the abnormal item does “ice-bearing”

A/ Abnormal item moves far less than the normal items

The moving objects are represented by rectangles.
In the next video, the simulated movements can be viewed.

After processing the images using different sampling sizes, we find these composite images. The abnormal element is clearly visible and identifiable as the darker area.
The larger the sample size, the larger the contrast.

Click here to open the notebook

B/ Abnormal item is moving

The same mechanisms have been used as for case A/
The results also show a clearly identifiable abnormal item as the gray area

Click here to open the notebook

Data Source

Simulated data

Copyright

© 2019 Jean Pierre Tollenboom

YOU'RE IN GOOD COMPANY