POPULATION ESTIMATES BY HUMANS AND MACHINES

A COMPARISON OF POPULATION ESTIMATES OF SMALL URBAN SETTLEMENTS DERIVED FROM AERIAL PHOTOGRPAHS BY HUMAN AND MACHINE INTERPRETATION INDICATES THAT THE ACCURACCY AT ANY GIVEN CENTRAL PLACE HAS NOT BEEN SIGNIFICANTLY IMPROVED OVER THAT OF DELIMITING BUILT-UP AREAS WITH IMAGE INTERPRETERS. AN IMAGE DISCRIMINATION, ENHANCEMENT AND COMBINATION SYSTEM (IDECS), A SYSTEM DESIGNED FOR IMAGE ENHANCEMENT AND INFORMATION PROCESSING, WAS DEVELOPED FOR INTERPRETING AND MEASURING THE SIZE OF URBAN AREAS FROM WHICH POPULATION ESTIMATES CAN BE COMPUTED QUICKLY, AND A COMPUTER PROGRAM WAS PREPARED FOR THE PURPOSE. REGRESSION STUDIES INDICATE THE ACCURACIES OF SEVERAL EXAMPLES BASED ON BLACK-AND-WHITE AERIAL PHOTOGRAPHS. MACHINE INTERPRETATION OFFERS SAVINGS IN TIME OVER HUMAN INTEPRETATION. THE SMALL ERRORS IN OBTAINING TOTAL POPULATION WITH EITHER HUMAN OR MACHINE PROCEDURE DEMONSTRATES THAT THE RELATIONSHIP EXISTING BETWEEN POPULATION OF URBAN PLACES AND THEIR RESPECTIVE URBAN AREAS COULD BE USEFUL TO GOVERNMENTS OF ALL NATIONS. /AUTHOR/