Multisensor data fusion for surface land-mine detection

摘要

Receiver operating characteristic (ROC) curves have been used tonexamine a novel target recognition system using a number ofnknowledge-based techniques to automatically detect surface land minesnthat are present in 30 sets of thermal and multispectral images. Ansummary of the results, graphed at a probability of detection greaternthan or equal to 96%, shows the false-alarm rates (FARs) obtained usingnvarious combinations of fusing sensors and neural classifiers averagednover the 30 images. The results show that using two neural-networknclassifiers on the input textural and spectral characteristics ofnselected multispectral bands, we obtained FARs of approximately 3%.nUsing polarization-resolved images only, we obtained FARs of 1.15%.nFusing the best classifier output with the polarization-resolved images,nwe obtained FARs as low as 0.023%. This result has shown the largenimprovement gained in the sensor fusion. Also, an improvement is shownnby comparing these results with those reported in an existing commercialnsystem