Image data compression for interactive applications

A compression method and apparatus are provided for use in image data processing where two-dimensional pixel images have respective absolute depth values specified per pixel, and that depth data is required to be compressed prior to transmission or storage with the image data. The process of compressing the depth data per image begins by determining patches of adjoining pixels of the image (90), which pixels of a patch have the same or similar depth values, and assigning a common depth value to all pixels of the patch. Next, pixel patches having the same or similar common pixel depth values are grouped together (92), and a common depth value is assigned to all pixels of the patches so grouped, following which for each discrete pixel depth value, a determination is made (94) as to the area or number of pixels of the image having that depth value. For the smallest of these areas, the depth value of each of those pixels concerned is modified (100) to the nearest existing discrete depth value. The determination of smallest area and consequent reassignment of depth value is then repeated until only a predetermined number N (for example 16 or 32) of discrete depth values remain. The reassignment of depth values with reference to the depths of adjoining pixels means that the depth data reduced to the N discrete levels may then be run-length coded to achieve a substantial reduction in depth data volume.