A Content-Based Method for Estimating the Slant Angle of Document Image

Abstract

The invention relates to an electronic processing category of a document, and provides a document image slant angle estimation method based on content. The method comprises the following steps of: firstly, carrying out binarization processing on an acquired document image; then, acquiring each linear segment in the image by a linear segment detection method, when judging that linear segment does not exist in the image, acquiring the linear segment by respectively positioning a position of a center line of each text line by a text line detecting algorithm; meanwhile, respectively counting distribution condition of a slant angle of each linear segment by a vote algorithm, and counting the slant angle of each linear segment by a mutation signal detection method of Gaussian wavelet transform; and finally, counting the slant angle of the document by occupied weight of the slant angle of each linear segment in the slant angle of the document image. According to the document image slant angle estimation method based on the content, the slant angle of the document image is estimated by the linear segment or the text line, not only the slant angle of the document image of the linear segment can be estimated, and but also the slant angle of the document image in which the linear segment does not exist can be estimated, so that the generality is good, the stability is good, and the precision is high.

Description

technical field [0001] The invention relates to the field of document electronic measurement and processing, in particular to an estimation method for document image tilt angle measurement, in particular to a content-based document image tilt angle estimation method. Background technique [0002] As the carrier of information, documents occupy a very important position in social life. They can enter computers through scanners, digital cameras or document processing systems, and convert them into document images or electronic documents, so that people can conveniently and effectively process them. Storage, management, transmission. [0003] In real life, due to the mechanical error of paper-feeding equipment such as scanners or the influence of human factors, the acquired document image usually has a certain degree of inclination. However, the processing object required by the document processing system is a neat document image, or a document image with a known tilt angle, o...

AI Technical Summary

Problems solved by technology

[0006] However, the calculation accuracy of the tilt correction method used in most image recognition technologies is largely affected by the image texture, and there are many types of documents and the layout is very complex, including text, tables, images, graphics, etc., and the general document image tilt Angle estimation methods are difficult to establish

[0007] A method for estimating the tilt angle of a document image is published in the document "Content-Based Document Image Slant Correction", but it uses length smoothing preprocessing for the document image, so that the text lines are connected into a connected area. Only the straight line segment corresponding to the table can be detected in the segment, and the versatility is not strong; and the longest line segment in the document is selected as the effective straight line segment to estimate the inclination angle of the document image, which ensures the calculation speed but reduces its accuracy

Images