Abstract: A method and apparatus for compressing digital data for transmission and reception in a narrow bandwidth, the method comprising preliminarily smoothing raw data according to a spatial convolution or running average technique, sampling the convoluted or averaged data to produce a subset having information from the full set of raw data, but fewer data points, determining the energy of the subset with a transform coder, encoding the transformed image, predicting the full set of raw data points at and between the points in the subset by means of a prediction algorithm, subtracting the raw data from the thus-predicted values to develop a difference signal, encoding one field of the difference signal and transmitting the encoded transformed data and the encoded difference signal to a receiver employing the same prediction algorithm. The signal is decoded in the receiver to produce both the transformed image and the difference signal.