Overall Shape Analysis-based Word Recognition Technology And Its Applications

Optical character recognition has brought forward many important studies and practical application of image procession and pattern recognition. License plate recognition is one important application of optical character recognition.For the impact of degraded images and small size of the state region in the system of American license plate recognition, it's quite difficult to segment characters from the images, which is required for traditional character recognition method but not needed for word integrated recognition. For the small number of American states, state recognition based on small lexicon is feasible.In this dissertation, the methods of image enhancement in practical application are analyzed including space domain filter, intensity stretching and histogram equalization. In combination of maximum extension feature and all-direction grads feature, a shape analysis based word recognition approach is brought forward to recognize the plate state for American license plate recognition system. Through experiments, it proves to be an effective character recognition method for the degraded and burred images.