Subjective quality of visual telephone pictures impaired by video crosstalk

Quality-grading tests have been made to determine quantitatively the effect of video crosstalk in a visual-telephone system with a scanning standard of 319 lines/picture, 50 fields/s and a nominal bandwidth of 1 MHz. The wanted and the disturbing signals were of types likely to be encountered in the local line network and were in analogue form. Although the wanted pictures were head-and-shoulders views of typical visual telephone users, two disturbing picture signals, one a head-and-shoulders view, and the other an electronically-generated caption, were used to represent the extremes of the probable range of subjective impairments. Three crosstalk paths were investigated, representing far end, near end and flat or undistorted crosstalk. By defining the crosstalk path at a frequency of 400 kHz, a single value of effective crosstalk attenuation, corresponding to a particular level of subjective impairment and a particular type of disturbing signal, is applicable for both far-end and near-end crosstalk paths. For a subjective impairment of ¿ imp, this value is 33 dB for a head-and-shoulders disturbing signal and 40.5 dB for a caption disturbing signal. The corresponding values of signal/crosstalk ratio for flat crosstalk are 42 dB and 43.4 dB.