Imagery data compression mechanism employing adaptive vector quantizer

Abstract

Imagery data is compressed by subdividing the data into a plurality of blocks. Each block is encoded into a bit map and a pair of threshold values. The bit map provides an array of binary encoded imagery data-representative vector combinations. A vector combination is subdivided into a set of adjacent sub-blocks, each which is examined to determine whether or not it is associated with a prescribed attribute of the image. If the vector is associated with a prescribed attribute, a code representative of that attribute is generated. If the vector is not associated with a prescribed attribute of the image, prescribed portions of the vector are inverted to derive modified vectors that are associated with prescribed image attributes. For each such modified vector a corresponding code is generated. Using each of these corresponding codes, image data is effectively reconstructed and then compared with the original image data. That modified vector which results in the minimum error between the reconstructed image data and the original data is then used to generate a prescribed binary output code representative of an associated attribute of the image. The binary output codes of a respective block of imagery data are then assembled and, together with the threshold values for that block of data, transmitted as a combined output code. At the receiver site, the image data is reconstructed using a look-up table mechanism in accordance with a translated bit map and the transmitted threshold values.