An improved edge-based text region segmentation algorithm applied to slab image data from steel plant
SungHoo Choi, Jong Pil Yun, Boyeul Seo, Jeehoon Park, Keunhwi Koo, Jonghyun Choi, Sang Woo Kim · 2008
This paper describes about the extraction of the text regions. In steel making process, it is very important to recognize the product management numbers because workers use it to tell quality and uses of slabs. Therefore automatic extraction of the slab management numbers is a prerequisite for automatic character segmentation and recognition system. In these days, many researchers have studied about the automatic text region detection from natural scenes, video frames, etc. However, these algorithms are not well suited for the steel plant environment. In this paper, the detail process of an improved edge-based horizontal text region detection method is described. Based on the above process, the selective binarization and the iterative character estimation are performed to find both side boundaries of the text region. Experimental results show that proposed method is robust and has higher detection rate.