ADVANCED PARTICLE FILTERING FOR MULTIPLE OBJECT TRACKING IN DYNAMIC FLUORESCENCE MICROSCOPY IMAGES

Quantitative analysis of dynamical processes in living cells by means of fluorescence microscopy imaging requires tracking of hundreds of bright spots in noisy image sequences. Deterministic approaches that perform object detection prior to tracking usually produce many incorrect tracks. We propose an improved, completely automatic tracker, built in a Bayesian probabilistic framework. It fully exploits spatiotemporal information and prior knowledge, yielding more robust tracking also in case of photobleaching and object interaction. Results from a preliminary quantitative evaluation based on highly realistic synthetic image sequences as well as real fluorescence microscopy image data in comparison with manual tracking indicate superior performance.