Automatic detection of image impairments

Abstract

Given only the sampled video image as input, methods to detect automatically image impairments that are visible to the eye are investigated by means of signal processing. The generalized classical cepstral approach is discussed, and thereafter specific investigations of the impairments. Sharpness of focus is measured using rise time properties of edges in the time domain. Echoes or ghosting are identified by investigating the power of the output of a variable correction filter. Top and bottom level signal saturation (also known as white level burning and black level crushing) is detected by grey level histograms.<>