Fast mean shift based traffic image filtering algorithm

Abstract

This paper describes a novel fast mean shift algorithm based on an accelerated iteration strategy. This new method focuses on solving the problem of high calculation complexity when high data dimension or large data sets are involved in mean shift. By predicting the mean shift vector, improved method reduces the number of iteration and speed up the calculation. The application of traffic image filtering is provided also. Experiment results of traffic image filtering demonstrate the efficiency of the fast mean shift algorithm.