Some computer experiments in picture processing for data compaction

Abstract

This paper presents two digital methods for data compression of gray tone pictures of human faces by dividing the picture into many subpictures and transmitting the statistics of the subpictures, assuming a uniform or truncated gaussian distribution for the video values. First, a method of generating the video values (samples) of each subpicture at the receiving end is given. Then a scheme is presented to arrange these samples in the subpicture by making use of the differences between the means of video values of the subpicture and the neighboring subpictures. Also incorporated in the method is a feature to copy some subpictures from the input picture into the output picture, which have a wide variation of intensity levels, while keeping the data compression at 1 bit/pixel. Computer processed pictures are given as examples.