A target tracking algorithm based on the continuous wavelet transform of spatio-temporal processing block

A novel target tracking approach based on the continuous wavelet transform (CWT) is proposed. Three energy densities are used to estimate motion parameters by sequentially optimizing a state vector composed of velocity, position, and size parameters. The practical functional relation is created to reasonably obtain the scale parameter. For the purpose of lowering computational cost, the spatio-temporal processing block for each frame is adopted as the tracking strategy. Both results in the simulation and in the real video data show that our approach is effective and has the advantages of being robust to local illumination variations as well as temporary partial occlusion by processing all the image windows in every built spatio-temporal processing block simultaneously during the tracking.