Engineering Physics Annotation << Back METROLOGY OF SPECTRAL SELECTION OF OBJECTS IN MULTI-SPECTRAL IMAGES T.Yu. SAGDULLAEV, Yu.S. SAGDULLAEV This article discusses issues related to the assessment of the metrological accuracy of spectral selection of objects in multi-spectral images. The information properties of multi-spectral television and thermal imaging images are determined by the selected spectral regions, the number of registration areas for the radiant flux, and the conditions for observing objects. A structural diagram of the spectral selection of optical images is shown, which takes into account interfering factors in both the optical and electrical paths, which to some extent affect the potential quality of the multi-spectral images generated for analysis. For one of the evaluations of the informativity of the multi-spectral images, the concept of a potential information vector of an object is introduced, which characterizes its difference from the others by the spectral portrait of the object and the number of delivered discriminative information for the consumer. Based on this, an approach to the metrological assessment of the accuracy of the selection of objects from the magnitude of interfering factors (interference) according to their spectral characteristics in one or several areas of registration of the radiant flux is considered. The dependences are shown that reflect a decrease in the accuracy of the spectral selection of objects with an increase in the magnitude of the error and the reliability of determining the true distribution of the spectral and energy characteristics in a given spectral region. The same sign in the error increment, when determining the true value of the information vector of an object, will introduce a smaller error compared to when for each object {N} in the «wide» spectral region the error from zone to registration zone will change its sign, that is, take values either or (+ΔIi ),or (–ΔIi ). Key words: discriminative information, information content of multi-spectral images, television and thermal imaging systems, evaluation of information content. DOI: 10.25791/infizik.09.2019.839 Contacts: - Pp. 44-50.