Object-oriented coding for efficient transmission of space-acquired image data

Object-oriented coding is introduced and compared with cosine transform coding for the transmission of space-acquired image data. The new coding scheme improves the compression ratio by incorporating knowledge about the image contents and its source. Utilising a priori knowledge enables on-board spacecraft data analysis to only select useful information for transmission. Initially, the cosine transform compression technique is reviewed and its merits illustrated. It is shown in a qualitative way by image examples that this form of compression is very limited. Next, an image analysis and coding method that utilises a priori information is introduced. This technique segments objects from the background, detects the segmented objects and evaluates a set of feature vectors for all objects. The vector set can then be transmitted as an object-oriented image representation with a comparatively high compression ratio. (C) 1997 Elsevier Science B.V.