Confidence evaluation method for scene matching aided navigation system

A novel scene matching terminal evaluation method based on evidence theory is put forward for scene matching aided navigation system, to deal with potential fault output because of noise, cloud, camera jittering and so on. The spatial-temporal correlation of sequence frames is founded firstly, and the identification framework is constructed with the maximal peaks of the matching correlation matrix. Then the basic belief assignment of the peaks is calculated, and the discounting operation is processed using suitable-matching factors. The combination rule of evidence reasoning is adopted to fuse the data of sequence frames. At last, the matching confidence or error alarm is given according to the decision criterion. Meanwhile a PCR-DSmT combination rule is proposed to solve the illogical fusion of Dempster rule in high conflicted situation and DSmT in case of multi evidences' fusion. Experiments with aerial images and Google earth satellite image prove that the approach is feasible and effective.