Human vision based pre-processing for MPEG video compression uses light
intensity. An input video signal is input to a contrast gain control
circuit where a Gaussian pyramid is constructed. The reduced pyramid image
serves as a local light level. The reduced pyramid image also is processed
to obtain a global light center. The difference between the local light
level and the global light center is used as an index to a lookup table
that provides a local gain control signal. The local gain control signal
and a global gain control signal input to the contrast gain control
circuit are used to multiply a Laplacian image derived from the input
image, the resulting modified Laplacian image being subtracted from an
equivalent lowpass filtered video image to produce a pre-processed output
video signal having reduced bandwidth for input to an MPEG2 compressor.
