Iterative predictor-based detection of image frame locations

摘要

Locations of respective image frames contained on an image recording, such as a continuous color photographic film strip scanned by digitizing opto-electronic scanner are identified by storing scanline data produced by the scanner in a digital database, and processing the stored scanline data in accordance with set of image frame identification operators, which iteratively identify locations of nominally valid frames, beginning with the identification of all well formed frames. Each well formed frame has prescribed image frame attributes including at least a spatial region of image modulance bounded by leading and trailing edges adjacent to Dmin interframe gaps. The iterative identification procedure includes 'chopping' less than well formed frames, sorting frame regions based upon geometry considerations and identifying and adjusting the size of oversized and undersize frames. To reduce computational complexity a set of image frame-based 'predictors' and an associated set of thresholds are used in the iterative search and sort process.