LINEAR PREDICTION IN TELEVISION 



777 



0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 



^1,0/^0,0 



Fig. 13 — Power reduction for various weighting coefficients for the previous 

 horizontal sample. 



areas as can be observed in the background of Scene A. Where the 

 separation of a white to black area is made, the error signal is large. It 

 is tliis large error signal that informs the receiver of this change in bright- 

 ness, and until another change occurs, the error output is again zero. This 

 type of performance produced the flat grey appearance of the back- 

 ground. In this way, the picture represents only changes in brightness — 

 a first difference type of picture. 



It may be noted that horizontal contour lines have vanished leaving 

 only vertical contours which pertain to the brightness changes that 

 have occurred. This effect is especially evident in Scene C. The 

 power reductions given in the lower left hand corner of these pictures 

 are consistent with their complexity. 



Fig. 16 shows the error signal appearance for "slope" prediction. 

 When compared to the error signal for "previous value" prediction 

 a finer vertical granularity is observed, and this is attributed to sudden 



