Mutilayer perceptrons model for image quality assessment

Autor: D. Dordevic, Dragan Kukolj, M. Drazic, Maja Pokric, I. Ostojic
Rok vydání: 2013
Předmět:
Zdroj: 2013 21st Telecommunications Forum Telfor (TELFOR).
DOI: 10.1109/telfor.2013.6716342
Popis: This paper presents an evaluation of the image quality using a nonlinear model based on Neural Network, more accurate Multilayer Perceptron. The main objective of estimation is a subjective rating of image quality with the help of the so-called Mean Opinion Score (MOS) scale using different combinations of input parameters which present objective metric for assessing the quality of multimedia content. In order to test the performance of the model and identified the how well the proposed model estimates the MOS the reference linear model (linear regression) is chosen. In addition, the paper examines how the selection of input data affects the quality of estimation MOS.
Databáze: OpenAIRE