Efficient matrix factorization for digital image processing. Informal report
C Moler, G. W. Stewart · 1979
Matrices obtained by digitizing photographs and video images can easily be too large to fit into the main memory of modern computers. This report describes a matrix factorization, the PDQ factorization, that requires much less arithmetic than previous factorization methods and, perhaps more important, does not require the matrix to be stored in the main memory. The emphasis is on the application of PDQ to image representation, but its applicability to image restoration and image enhancement are also addressed. 3 figures.