Comparison of the mid-frequency line edge noise algorithms of Jim Hamerly and Yigal Gur for the best correlation to the psychophysical defect known as raggedness

Abstract

A study was conducted to compare the mid-frequency line edge noise algorithms of Jim Hamerly and Yigal Gur. Comparisons were made to determine which algorithm correlated best with the psychophysical defect known as raggedness. Tests were conducted on "standard" observers using a series of test images with varying amounts of edge noise. A master image was produced photographically and ensuing images generated with random error using electrophotographic processes. Scanning of the images to determine output metric values was done on a Zeiss microdensitometer. It was found that with electrophotographic prints, the Hamerly algorithm has a correlation coefficient of .83. Determination of metric values for the Gur algorithm were not possible due to an error in scanning procedure. A subjective evaluation of his algorithm is included in its place.