Slips Of Character Extraction And Recognition Research

China’s ancient culture is profound and profound.Among them,ancient Chinese characters are the carrier of our understanding of Chinese history,politics,economy and humanistic customs.At the beginning of the bamboo slips is the most important text carrier,until the emergence of paper,before the end of the text as a carrier of history,so bamboo slips of national culture inheritance and development of great significance.The text of the slips has many characteristics,such as different characters,local deformation and a lot of noise.It is not only a lot of manpower and material resources,but also slow progress,so the use of advanced modern technology,digital image processing,pattern recognition technology The research on the extraction and recognition of slips and phrases is of great significance.In this paper,digital image processing and pattern recognition as a theoretical guide,first of the bamboo text image samples gray,binarized,denoising,refinement,anti-color and normalization and other pre-processing,and image denoising A new adaptive weighted median filtering algorithm is proposed.The method of fractional order differential is used to detect the image noise.The noise-free image is used to determine the size of the filter window according to the number of noise points,and the weight is calculated by adaptive.Point weighted median filter.The results can achieve better de-noising effect and retain more text details,is conducive to follow-up text recognition.And then the feature extraction and feature extraction are the indispensable steps in the character recognition.This paper presents a method to fuse the three characteristics of the double elastic local density and the generalized generalized density and the structural features of the component,which makes up the deficiency of the single feature,Better meet the high stability,accuracy.Finally,the recognition of the slips in the standard BP neural network and the improved BP neural network is carried out.The experimental results show that the multi-feature fusion has a high recognition rate and also proves the validity of the multi-feature fusion.