Image Morphology and Statistical Analysis for Number Plate Recognition

Two algorithms are proposed: Edge Map Statistics (EMS) and the Multiple Morphological Filtrations (MMF). The EMS might approximate the Hough transform and it was applied on the edge map after connected component analysis. However, it could not be adopted because it had an inadequate accuracy and did not respond very well to images of differing illumination. The MMF in the other hand is based on a set of morphological operations that are applied to the edge map. The irrelevant objects are detected by their location and connectivity information; leaving the plate region as the highest contrast object from the remaining components. Around 93% of success rate was achieved using MMF approach.