Adaptive differential pulse code modulation video encoder

Abstract

An adaptive DCPM video encoder operates directly upon a digitized composite video signal to generate a bit reduced serial digital output signal. The encoder has a feed back loop in the form of a recursive filter with zero's at d.c. and the color subcarrier frequency of the digitized encoded video signal that determines a predicted value for subtraction from the digitized composite video signal to produce a difference signal having the color subcarrier frequency nulled out. The difference signal is quantized to produce a quantized difference signal from which the predicted value is derived. An inner loop determines from the quantized difference signal an offset value from which the presence of large step transitions in the digitized encoded video signal are detected. For large step transitions the offset signal is subtracted from the difference signal to produce an offset difference signal for input to the quantizer. All adders and subtractors use modulo summing techniques to reduce the size of their outputs by one bit. The quantization transfer function for the quantizer is non-linear, having fine resolution near zero and coarse resolution away from zero. A plurality of quantization transfer functions are selectable according to the size of the step transitions as determined from the quantized difference signal. For digitized composite video signal samples near the maximum and minimum values when wrap around occurs due to quantization errors, a clip logic circuit selects a clipped quantized difference signal for output from the quantizer.