Image Processing Applied to Optical Surveillance for Safeguards

The technique used until now for optical surveillance in safeguards generates a high amount of data but only a small part is safeguards relevant. Image processing, which involves the extraction of useful features from image by computer, can perform a data reduction either for on-site surveillance or for reviewing of stored data. The developed prototype detects scene changes in zones of interest, stores these images and can evidence the scene modifications in the picture. The used method based on a cross-correlation function between observed and reference image, is to a large extent, insensible to illumination variations and sensor noise. In its present state the application of the system is limited to situations with few scene changes in the zones of interest (f.i. storage areas). Some tests in this application field have given encouraging results.