Tracking Multiple Fish in a Single Tank Using an Improved Particle Filter

摘要

Studies on tracking fishes have become a popular research endeavour in recent years. Many methods have been used to track fishes by integrating microchips in fishes, using infra-red cameras, image processing and motion sensor. The use of particle filter in the process of tracking has been widely used by researchers. Particle filters is used to track people, fluid movement and animals. In this paper, the particle filter algorithm is improved to track multiple fish in a fish tank. The aim is to identify every fish trajectories and fish target location for further analysis. The main challenge is to ensure that the correct fish are tracked and the algorithm manages to identify specific fish even if they overlaps with each another. The objective of the study is to improve the existing particle filter to track multiple fish in a single fish tank. The improved algorithm contains an additional cache which stores the object's position to estimate the next potential move of the fish. The result is evaluated by comparing existing algorithm without the enhancement with the improved algorithm. Besides, suggestions in improving the particle filter will also be discussed in this paper.