Evaluation of image noise using psychophysical RMS granularity

摘要

The psychophysical RMS (root mean square) granularityσ{sub}p was proposed for the evaluation of image noise. The value ofσ{sub}p was calculated from the Wiener spectrum of the image noise and the frequency dependence of the minimum perceptiblecontrast. The minimum perceptible contrast expresses psychophysical properties of human vision. Its frequency dependence which we determined was used. Examples were presented for three typical X-ray imaging systems. For each system, physical values ofSelwyn granularity coefficient S and our psychophysical RMS values ofσ{sub}p of noise patterns in samples having five uniform densities were respectively arranged in descending orders. The two orders of the density were compared with the subjectiverating results arranged in descending order of the perceptibility of the noise. The density dependence ofσ{sub}p was in close agreement with the subjective rating results that (I) the most perceptible density was 1.0 and (2) the perceptibility at density 2.5 was much worse than that at 0.6. Whereas the physical value S did not correlate with the results of subjective rating.