Identification And Application Of Hand Filled Paper Form

The intelligent identification and analysis of paper forms are widely used in the digitization of personnel files,personnel assessment,questionnaire survey and so on.Starting from the business requirements of the personnel assessment subsystem,this thesis studies a form recognition method based on template and positioning technology,and uses this technology in the personnel assessment subsystem to realize fast and efficient form data recognition and input.This technology is applied to the scene where the table template is known.The table structure information is obtained by labeling the template image,and the table structure is stored in the format of convenient input database table.For the table image to be recognized,the template type is recognized through the anchor point,and the position of each cell is preliminarily obtained.Then the table image is corrected,preprocessed and the table text box is positioned.Finally,the recognition result with the image structure information is obtained through text recognition and displayed on the web page.The identification results can be stored in the database through a small amount of manual error correction to meet the application requirements.The main work of this paper is as follows:(1)Table structure information representation.A form structure information representation method based on template is designed.Different regions in the form are given different types,and the information of each region is stored in tree structure.The unique meaning area of the table is stored in the root node of the tree,and the repeated meaning area is stored in the leaf node.A table is a tree,and a piece of data is a branch of the tree,so that the corresponding relationship between the table data and the database table fields can be established to facilitate the subsequent data entry into the database and improve the entry efficiency.(2)Table recognition based on template and location technology.A table recognition method based on template and positioning technology is designed.After correcting and preprocessing the table image,the first positioning is completed by using the cell positioning technology,and on this basis,the second positioning of the text box is completed by using the image clipping technology.The position information obtained from the secondary positioning is aggregated and judged by intersection over Union(IOU)with the coordinates of the template cell,so as to accurately locate the position of the table text box while retaining the structural information of the table to the greatest extent.According to different recognition contents,specific data sets are used to train the model to improve the accuracy of model recognition.(3)Personnel assessment subsystem.Completed the implementation of the personnel assessment subsystem of the intelligent organization platform,realized the manual form identification and entry module based on the form identification method proposed in this paper,and improved the assessment process.After the table data is identified,it can be put into the database after a small amount of manual correction,and then the data in the database can be managed.