Rapid content-based retrieval from document image databases

A new method of applying n-tuple recognition techniques to searching document images is reported. The method involves scanning an n-tuple classifier over a chain-code of the image. In scanning n-tuple systems, the traditional advantages of n-tuple recognition, i.e. training and recognition speed are retained, while offering superior recognition accuracy. In this paper the application of the system to searching databases of comic-strip images is reported. The work is in its preliminary stage, and results so far while promising are based on too little data to draw any firm conclusions. However, the method is extremely efficient, and certainly warrants further investigation