Abstract: A method and apparatus for processing data using data compression. During compression, a relationship between a next data word and a previous data word is determined and that relationship is encoded into variable bit-length code words, the bit-length being inversely proportional to the probability of occurrence of the relationship between the data words. The encoded relationships are transmitted rather than the data words themselves. In one embodiment, a look-up table of code words is stored which is indexed according to the possible relationships. A relationship is determined between the previous and next data words and the relationship is matched to the index to find the corresponding code word in the look-up table. The corresponding code word is retrieved from the look-up table, is substituted for the relationship, and is transmitted.

Abstract: A computationally fast, effective method for compressing digital images uses an optimized subsampling process. A set of spatially overlapping spline functions is used as a basis onto which the image data are projected by use of a least-mean-squares criterion. The specified processes can be implemented for compressing and interpolating digital data arrays of N dimensions. Linear, planar and hyperplanar spline functions allow convenient, fast and efficient closed-form optimal compression, which process is easily incorporated into existing digital processing systems. A key advantage of the disclosed method is the fast coding/reconstruction speed, because it involves only FFT or convolution types of processors.