An efficacious medical video segmentation approach with a moving camera

This paper presents a novel algorithm for segmenting moving objects from a moving background scene that contains complex camera motion using gray or color medical videos. We develop a robust and efficiently background subtraction algorithm that is able to cope with local scene changes, as well as globe scene changes caused by camera motion. The algorithm is based on a proposed multi-panorama reconstruction method, which partitions the medical video sequence into a number of segments, for which independent panoramas are synthesized. The background subtraction in our method, which introduces risk maps to account for the problem that the background image may not be perfectly aligned to the input images, greatly improves the foreground separation even on the object boundary. We have applied this method to medical image sequences. The results, which demonstrate the system´s performance, and some refinement techniques we employed in our implementation are also shown.