Skip to Main Content
Wavelet packet analysis method is appropriate to process nature texture signal and a hidden Markov model has good learning interpretability and needs only small training samples. A wavelet packet-HMM-based method on road surface state recognition was proposed. The wavelet packet analysis was adapted to extract characteristic entropies from the image signals. Thus, four kinds of data on road surface were trained respectively to get the HMM to identify road surface states. The result of experiments shows that the means is effective.