Mean-shift visual tracking with NP-Windows density estimates

Abstract

The mean-shift algorithm is a robust and easy method of finding local extrema in the density distribution of a data set. It has been used successfully for visual tracking in which the target is modelled using a colour histogram, and the image window with best matching histogram is sought. However a histogram is potentially a poor estimate of the underlying colour distribution: it is not invariant to the image scale, the number of histogram bins or the number of samples, and this can have an adverse affect on the speed and accuracy of convergence of the mean-shift algorithm. We apply a general non-parametric PDF estimation method [7] to replace the histogram in mean-shift tracking to improve its accuracy. This algorithm uses an interpolation scheme which fits piecewise functions to the signal samples, calculating the PDF by accumulating the contribution of each piecewise function. Its accuracy is dependent only on the accuracy of the piecewise representation, not the number of samples or the number of bins. Experiments are conducted to demonstrate that we improve mean-shift visual tracking both in accuracy and speed.