Models of criterion evaluation of the image processing systems effectiveness

Abstract

The criteria used to evaluate the effectiveness of image processing systems are investigated in the article. The requirements for performance criteria are analyzed. Private criteria which are used for image processing systems are selected and justified. Such parameters as performance, system cost, hardware costs characterize the system itself and depend on its specific type. It was shown that the information capacity, the probability of completing the task, and the accuracy of the image processing algorithm are the main criteria that characterize the quality of the processing method. It is shown that such a reliability criterion as normalized root mean square error best meets the requirements of efficiency criteria. Criteria models that are based on the normalized root mean square error in relation to discrete images have been studied. The simulation results and obtained dependences of cost functions on the speed of system information processing are given. The methodology for choosing a generalized criterion, which characterizes not only the information processing system, but also the methods used to implement this system was found. We obtained a generalized cost criterion, which arguments are the accuracy of system operation, speed of operation, and cost advantages.