This article is based on the assessment of the quality\nof video signals, specifically an objective evaluation of completely\nreferenced video signals in standard definition. The most reliable\nway to measure the difference in quality between two video\nscenes is using a panel made from television viewers, resulting in\na subjective measure of the difference in quality. This\nmethodology requires a long period and has an elevated\noperational cost; this makes it an unpractical method to be used.\nThis article will present the relevant aspects for the assessment of\nvideo application in standard definition digital television and the\nvalidation of these methodologies. The objective is to test metrics\nbelow the computational cost that evaluate the peak signal-to noise\nratio (PSNR) and measures the structural similarity index\nmeasure (SSIM). One methodology for the validation of these\nmetrics is presented and is based on the scenes and the results of\nsubjective tests performed by VQEG. The scenes for these\nmetrics are prepared by the equalization of brightness, detail\nsmoothing, and edge detection. Controlling the intensity of these\nfilters, a new set of measures is obtained. Performance\ncomparisons are made between these new sets of measures and\nthe set of measures obtained by VQEG. The results showed that\nthe objective measures are easily implemented from the\ncomputational point of view, and can be used to compare the\nquality of video signals, if properly combined with techniques for\nthe adequacy of the human visual system like the mitigation and\nextraction of contours.
Loading....