Improving Video Segmentation Algorithms by Detection of and Adaption to Altered Illumination

Johan Lindström, Finn Lindgren, Ulla Holst and Kalle Åström

Centre for Mathematical Sciences
Mathematical Statistics
Lund Institute of Technology,
Lund University,

ISSN 1403-9338
Changing illumination constitutes a serious challenge for video segmentation algorithms, especially in outdoor scenes under cloudy conditions. Rapid illumination changes, e.g. caused by varying cloud cover, often cause existing  segmentation algorithms to erroneously classify large parts of the image as foreground.
Here a method that extends existing segmentation algorithms by detecting illumination changes using a CUSUM detector and adjusting the background model to conform with the new illumination is presented. The method is shown to work for two segmentation algorithms, and it is indicated how the method could be extended to other algorithms.