Removal of spurious edges from thinned character image: A survey
Nikita Khatri, Mukesh M. Goswami · 2016
Thinning is an important pre-processing step in many shape based character and word recognition systems. The one-pixel wide thinned character image is useful for identifying the shape of various strokes used to from the character, which is helpful in many applications such as word spotting in the document image, content-based document image retrieval, character and word classification, etc. However, all thinning algorithms generate unwanted spurious edges. As a result, the shape of the stroke is distorted which in turn limits the accuracy of stroke recognition in characters. Thus, removal of these spurious edges from thinned image will improve the state of the art in stroke recognition and benefit many applications stated above. This paper presents a survey of various techniques used for removal of spurious edges from thinned character image and tries to identify the challenges and research gaps in the problem.