The tradeoff between computational burden and detection accuracy is the real critical point of sliding windows, so the method of multistage particle windows is proposed. Firstly, a cascade structure of classifiers was established, then a set of corresponding particles were generated randomly in the whole image. Secondly, statistical-based search using a Monte Carlo sampling for estimating the likelihood density function was carried out by taking into account the feedback of the classifiers, and the proposal distribution of particles in multistage strategy until the coarse-to-fine detection is obtained. Lastly, this approach was applied to make object detection both in signal image and image sequence, then comparisons between the proposed approach and sliding windows indicate that the proposed method can provide higher detection rates and accuracy as well as a lower computational burden.
