BANDWITH COMPRESSION FOR FACSIMILE USING SIGNAL MODIFICATION.

Research on facsimile bandwidth compression in digital transmission has recently intensified. Most such research has been based on the removal of redundancy of sampled signals, with the result that these efforts have reached a fairly high level and any further substantial improvement of compression ratio is very difficult. This paper proposes a method which is based on a new approach. With consideration of the redundancy of the picture to be transmitted, the method utilizes the fact that even if the signals are locally modified from white to black or from black to white, picture quality can be adequate for practical use. This concept (signal modification) is applied to the two-dimensional prediction technique. The result is that the number of prediction error signals is reduced to almost one half and the compression ratio can be improved 20-60%. A prediction inversion method is proposed as a suitable prediction method for signal modification and it is shown that a further improvement of 10-20% in compression ratio is possible.