N. Barrón, J. Cruz, and L. Altamirano (Mexico)
Visual Tracking, Adaptive Sizing, MAP
In this paper, it is presented an adaptive approach for track ing that is strengthened with probabilistic methods to es timate the correct state of the object. Most of the tracking approaches are focused only on dynamic (temporal) behav iors of the target. However, this method is also adaptive to spatial characteristics of the target such as its size, due the size and the shape of the object can change over time. To estimate the target position, a Probabilistic Markov Model as Kalman Filter is used, and to estimate the object appear ance, the same method is used for defining a search space with random samples. Furthermore, the method maintains the tracking even when the target is affected by occlusion or noisy background. Tracking performance is tested with synthetic and real image sequences and we present preci sion and accuracy under different conditions.
Important Links:
Go Back
