Research On Character Refinement Algorithm Based On MATLAB

The skeleton is a simple shape description method that can maintain the topology and geometry of the original image.The skeleton extraction algorithms has been widely used in text refinement,medical image processing,robot path planning,fingerprint recognition,scientific image analysis.The rapid and accurate extraction of skeletons in the field of image processing is still a hot topic for researchers,and it has very important theoretical and practical significance.This article mainly studies how to achieve rapid and highperformance Chinese character refinement,and proposes a new type of connectivity principle.Based on this principle,a method of automatically generating a retained template is proposed.According to the new method,the OPTA series algorithms is improved,which eliminates the defects that breaks and burrs are more likely to occur in the series of algorithms,and greatly improves the efficiency of refinement operations.First,it studies the process of preprocessing of Chinese character recognition,at the same time,analysis the advantages and disadvantages of the most widely used refinement algorithms.The OPTA series refinement algorithm is a one-step iterative refinement algorithm that simultaneously iterates all pixels.The Z-S series refinement algorithm performs iterative operations in two steps and uses the logical operation formulas to process the target pixels.Through experimental comparison,it is found that there are poor skeleton connectivity,fine granularity,and slower granularity in the skeleton of the OPTA series refinement algorithms,and there is room for further improvement.Therefore,this paper does further research on these shortcomings.Then,it proposes a new type of connectivity principle,and on this basis,it proposes a method for automatically generating a retention template.This method is based on the delete template in the refinement algorithm to automatically match,and then find out the need to retain templates.we use this method to improve the OPTA series refinement algorithm and match the series of algorithms with simpler and more effective retention templates.The experiment proves that the refinement speed of the OPTA series algorithm has been greatly improved by using this method,and make up for some algorithm breakpoint defects,improve the connectivity of the skeletonIt proposes an improved OPTA refinement algorithm based on the OPTA series refinement algorithm and the new connectivity principle.The basic idea of the algorithm is to propose a new deletion template according to the new connectivity principle,and use this deletion template to automatically generate a retention template so that the skeleton has no breakpoints and is completely refined.Experimental results show that the algorithm can achieve complete refinement of Chinese characters without any breakpoints,and there are few burrs in the skeleton.Therefore,the refinement results have good connectivity and refinement.Finally,in this experiment,we find that the improved OPTA algorithm does not have the advantage of refinement speed,and the refinement of Chinese characters is too thorough,which results in the loss of some skeleton structure information.Therefore,it proposes an improved OPTA fast thinning algorithm.This algorithm also sets up the right adjacency and the next adjoining deletion template and retention template based on the new connectivity principle,and adds some small glitch removal templates and small area reservations.template.It also proposes a method to delete double slashes in the algorithm to ensure the single pixel of the skeleton.Experiments show that the algorithms refinement speed has obvious advantages compared with other refinement algorithms,and the refined skeleton also maintains the original graphs topology,refinement,and connectivity.The two improved refinement algorithms are experimented on the MATLAB platform.The datasets in Arial,Regular script,and Official script all have good refinement results and the algorithm is a more practical refinement algorithm.