Image ability to obtain correct matching based on feature matching
Jin Xia · Infrared and Laser Engineering · 2005
The evaluating approach of images′ ability to obtain correct matching based on image feature matching is discussed.It is suggested that information entropy and summation of image gradient can be used for evaluating images′ ability to obtain correct matching.By the test of image matching based on corner feature and Hausdorff distance, the relation among correct matching probability and in- formation entropy and summation of image gradient is discovered.Strong correlation exists among correct matching probability and information entropy and summation of image gradient.It is shown that correctmatching probability is increasing with the increasing of information entropy and summation of imagegradient, although concrete rule is of tiny difference because of the field surface landscape′s difference.Therefore, through calculating image′s information content, automatic reference image selection and plan-ning navigation route can be done.During real-time matching navigation, whether matching carried out ornot would be confirmed by evaluation real-time image′s information content.To do as such,higher correctmatching probability would be obtained, misguide would be avoided and matching time would be saved.The method of the paper can also be available in image evaluation based on gray-level image matching.