Picture table content extraction method based on computer vision and natural language processing

Abstract

The invention discloses a picture table content extraction method based on computer vision and natural language processing. The method comprises the following steps: 1, inputting a picture into a table frame identification model, identifying a table frame, and calculating coordinates of each cell in a table; 2, extracting the text content of each cell; 3, according to the extracted text content, labeling according to three types of keys, values and mixed values, constructing a table content classification data set, and training a cell content classification model based on the data set; 4, inferring a table layout according to the table coordinates, the cell coordinates and the category of each cell text; and 5, organizing the data in the table in a JSON format according to the layout information of the table, the content of the table cells and the category information. A natural language processing technology is introduced, the category of the content of each cell in the table is marked, the table layout is reasoned in combination with the position information of the cells, and finally the table content is output in a structured mode.

Description

technical field [0001] The invention relates to the technical field of table data extraction, in particular to a method for extracting table content from pictures based on computer vision and natural language processing. Background technique [0002] The application of information extraction based on computer vision and natural language processing technology is becoming more and more widespread, such as recognizing text from pictures, extracting entities such as names, place names, and phone numbers from text, and extracting key information from invoices, insurance policies, and other forms. Wait. At the same time, major cloud vendors also provide identification services based on cloud platforms for form data such as bills and contracts. [0003] Existing techniques for extracting tabular data mainly focus on two aspects. First, through traditional image processing methods, such as erosion, expansion, edge detection, contour recognition, etc., first identify the table in t...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a table content extraction method based on computer vision and natural language processing, using border detection, OCR, text classification, etc. Technology, develop table border recognition, table content extraction, table content classification, table layout reasoning modules, realize data extraction from image tables and convert the extracted data into structured data in JSON format

Images