Intelligent Communication for Low Bit Rate Channels

In this paper, intelligent compression techniques to allow transmission of text and image data over low bit rate channels are proposed. For text, a compression technique for English language text is presented. Compression is carried out by deleting characters that are not necessary for comprehension by a human reader. Decompression is carried out on a sentence by sentence basis using a database consisting of words and English grammar information. The possible sentences are constructed and then the incorrect sentences are eliminated until only one possibility remains. When this technique is used in combination with a lossless compression scheme such as Lempel–Ziv, the text can be compressed to a higher degree than is possible with only traditional compression techniques. For images, a technique to convert sequences of images into sequences of commands is described. Instead of transmitting the actual images, motion descriptions of the image relative to the previous image are sent. This results in high compression.