Morphological degradation models and their use in document image restoration

摘要: Document images undergo various degradation processes. Numerous models of these processes have been proposed in the literature. In this paper we propose a model-based restoration algorithm. The algorithm first estimates parameters model and then uses estimated to construct lookup table for restoring degraded image. is used estimate probability an ideal binary pattern, given noisy observed pattern. This by degrading noise-free document computing frequency corresponding pattern pairs. conditional restore images. impact process quantified decrease OCR word character error rate. We find that parameter values, decreases rate 16.1% 7.35%. some categories (e.g. give rise broken characters) there 41.5% reduction 20.4%