Research On Tracking Neurofilaments Based-on Particle Filtering

Neurofilaments are important part of neuron axon,which are fiber filament composed of neural cells,and they have very important significance in supportment of cell structure, the maintenance of cell state and axonal transport etc. So by tracking the neurofilaments, to grasp its motion characteristics. At present, many biomedical tracking is carried out manually, for the tracking of large amount of data and images video frames,this method is very time-consuming and tracking error is big.This paper firstly introduces the particle filter algorithm and Kalman filter algorithm,then compares these two algorithms in the example, finally the operating results show that the particle filter algorithm has more advantages in target tracking.This paper selects the particle filter algorithm for neurofilament protein tracking,establishes the state model with the propagation of neurofilament protein,and then build the observation model with the kernel-based color histogram distribution, this two models eventually forms a tracking model for neurofilament protein.Finally,this paper uses Matlab Toolbox to simulate single and multiple neurofilament protein tracking respectively. Firstly, this paper selects 100 and 200 particles for single neurofilament protein tracking, the tracking result of 100 particles shows that in addition to the deviation of individual particles, other particles can track accurately;200 particles tracking result shows that all particles can accurately track, and tracking accuracy is better than 100 particles.Secondly,this paper finishes 100 and 200 particles tracking simulations for multiple neurofilament protein.The 100 particles tracking result shows that it can accurately track multiple targets with a good tracking accuracy;200 particles tracking result shows that the tracking accuracy is significantly improved compared to 100 particles. Compared with the traditional manual tracking methods, the particle filter algorithm not only improves the tracking precision and accuracy, but also saves a lot of computing time and improves the tracking efficiency.