An image data compression and decompression technique applies one or more compression code tables to optimally compress an image data stream. The compression code tables are established in accordance with anticipated image characteristics, and to accommodate different levels of variation or entropy in the image data. The image data may be divided into blocks or subregions for analysis of which of the candidate compression code tables provides the optimal compression of each subregion. The appropriate code table is selected for each subregion. The evaluation of the compression performance based upon application of each compression code table may include analysis of prediction differences or errors between predicted values for pixels of an image and the actual values for the pixels.