Image Skew Correction Of The Republican China Documents

Documents of Republican China are a kind of extremely valuable and important documents.Digitization by scanning will cause images to be tilted,which not only affects the visual effect of image retrieval,but also is not conducive to the further processing of document images.Through digital image processing technology,it summarizes the layout characteristics of the documents of Republican China images,defines the type of the documents of Republican China images and designs an adaptive image skew correction algorithm that automatically select the corresponding correction technology for different types of the documents of Republican China images,laying the foundation for practical application.The main work of the thesis includes the following three aspects.First,sort out and analyzed the layout characteristics of the paper documents of Republican China images,so that to defining the types of documents images of this period.According to the visual analysis of documents of Republican China images,the researcher found that these documents images are mainly composed of tables,text columns,and seals.These constituent elements provide guidance of skew correction of image.And according to the gray distribution analysis and spatial correlation analysis of documents of Republican China images,the researcher also found that the texture-based tilted correction technology is not suitable for processing the documents of Republican China images.Based on the layout structure analysis and image feature analysis,documents of Republican China images categories that suitable for correction is defined,they can be divided into three types: archival images including table lines,archival images mainly composed of text columns,and complex layouts.Second,design an adaptive skew correction algorithm for documents of Republican China images.The image type is determined according to the layout characteristics of the documents of Republican China image,and a targeted correction algorithm is used according to the image type.The entire process runs automatically,and the input image is adaptively processed without manual intervention.In this paper,six skew correction techniques are adopted to correct the documents of Republican China images including table lines,documents of Republican China images mainly composed of text columns,and documents of Republican China images that complex layouts.The average error is used as a measure.Simulation experiments show that documents images including table lines adopts Hough Transform to achieve the best effect,documents images mainly composed of text columns adopts the projection method to achieve the best effect,and the complex layout documents images adopt skew correction technology based on image moment to achieve the best effect.Third,the designed skew correction algorithm is used in a digital archive production management system.First of all,the paper analyzes and designs the application management system of archival production management of the Republican China,meanwhile sort out the performance index such as entropy(E),peak signal-to-noise ratio(PSNR),cross entropy(CE),normalized cross-correlation coefficient(NCCC)to compare the accuracy of images tilted correction,and they are used as the evaluation standard of the documents of Republican China image's correction performance.Next,through correcting the documents of Republican China images,to compare and analyze the various indexes of the paper's skew correction algorithm.Finally,the correction performance of the skew correction method proposed to the documents of Republican China images are proved.