Font Size: a A A

Research On Offline Hard Pen Chinese Character Stroke Extraction Technology Based On Point Set Matching

Posted on:2020-05-17Degree:MasterType:Thesis
Country:ChinaCandidate:Z F LiFull Text:PDF
GTID:2438330578977139Subject:The modern education technology
Abstract/Summary:PDF Full Text Request
In the face of the continuous decline of Chinese character writing level,more and more people need to practice handwriting,and at the same time need timely evaluation to correct the inadequacies in writing.However,due to the limitation of time and energy,it is very difficult for evaluators to make targeted guidance and judgment for all learners’ handwritten works.Therefore,the automatic evaluation of the quality of rival writing works becomes more and more important.Based on the in-depth study of Chinese character stroke extraction technology and combined with the design ideas and schemes of stroke extraction in regular script works,an offline hard-stroke extraction technology based on point set registration is designed.The technique uses binary image to fill holes and remove impurities,extract skeleton,segment skeleton,map skeleton point set and template point set,and combine segmentation and mapping results to extract corresponding strokes.Optimize the strategy and parameters of each step in the stroke extraction process according to the actual extraction effect.The focus of the work involves every step in the stroke extraction process.The ultimate goal is to accurately extract each stroke of the handwritten hard-stroke,so that the level of application can be reached.Compared with this kind of research problem,this paper creatively designs a new method of sub-region binarization which dynamically sets threshold values according to the characteristics of different regions,proposes a new method to fill holes and remove impurities based on region growth,and a new method for extracting strokes by combining segmentation with mapping results is designed.Based on the above research results,the success rate of stroke extraction is 69%,showing a good extraction effect,but there is still a certain gap with the practical application.This research also points out the direction for the follow-up research in the future.Look forward to further efforts by the researchers.
Keywords/Search Tags:pen writing, stroke extraction, point set mapping, matching
PDF Full Text Request
Related items