Rapid digitization method and system for meteorological paper form document

Abstract

The invention provides a rapid digitization method and system for a meteorological paper form document, and belongs to the field of meteorological data processing. The rapid digitization method comprises the following steps: acquiring an image of a meteorological paper form document; after the image is preprocessed, extracting table lines, carrying out character recognition, completing extraction and positioning of characters, and storing the characters as first-class characters; cutting the processed table document image, and carrying out character input on each sub-image; carrying out positioning matching and verification on the first type of characters by using the input characters; when the verification is passed, storing the characters; when the verification is not passed, discarding the characters, taking the current sub-image as a table document image, and turning to preprocessing; and when all the sub-images are traversed and verification is passed, outputting all the characters as digital results. According to the method, rapid digitization of the meteorological table document is realized, text detection based on semantic segmentation is simplified, and high-precision and high-accuracy character extraction of the table document containing printed and handwritten characters is realized.

Description

technical field [0001] The invention belongs to the field of meteorological data processing, and in particular relates to a rapid digitalization method and system for meteorological paper table documents. Background technique [0002] In meteorological observation, the initial observation results are meteorological images, and various localities collect local meteorological data through the digitization of meteorological images, and upload them to the meteorological center, and the meteorological center conducts statistics and analysis of all meteorological data. When collecting meteorological data in various places, due to the time series and multi-parameter nature of meteorological data, it is usually necessary to organize the data through forms; the arrangement of meteorological paper form documents is generally done by manual observation and handwriting. Then, through the identification of the paper form, for example, the OCR text recognition method in artificial intelli...

AI Technical Summary

Problems solved by technology

In image skew correction, good accuracy is currently not guaranteed for table document images with handwritten numbers

In the process of image binarization, the binarization method is relatively simple, and it is often unable to adapt to various conditions in practical applications, such as: uneven brightness of scanning results, shallow stroke grayscale, image histogram The distribution of gray value is relatively uniform, etc.

In the process of table character positioning, it includes using prior knowledge positioning and table line detection positioning; among them, the prior knowledge method needs prior knowledge and cannot automatically adapt to various forms of different formats; the table line detection method includes lateral projection method, contour extraction method, table line intersection analysis method, and Hough transform method; the side projection method is simple and fast, but it is more susceptible to the influence of table tilt and character sticking table lines; the contour extraction method is also easily affected by the table line Errors are caused by breaking or sticking table lines; the table line intersection analysis method iteratively builds all the cells in the table by classifying and matching adjacent intersections, but this method is time-consuming and susceptible to broken tables line effect

Images