Tibetan historical document text line segmentation method based on baseline estimation

Abstract

The invention relates to a Tibetan historical document text line segmentation method. The method comprises the steps that the image of the left part of a Tibetan historical document is extracted; Tibetan vowel sound nodes and certain prominent strokes are removed; the starting position information of the baselines of the Tibetan text lines and the number of text lines are acquired; according to the starting position of the baselines, the baselines are established from left to right; during the establishment of the baselines, the baselines needs to be dynamically adjusted according to the pixelvalues of the surrounding points; the estimated baselines are used, and a communication area analysis method is used to determine the position of an adhesion area from two baselines for segmentation;and finally text lines are separated. According to the invention, the Tibetan historical document text line segmentation method based on baseline estimation is more suitable for the segmentation ofthe text lines of the Tibetan historical document, and has more segmentation precision than a traditional technology based on projection segmentation; and compared with a method based on projection segmentation, the method has the advantage that the segmented text lines are more correct.

Description

technical field [0001] The invention relates to an image processing method, in particular to a text line segmentation method of a Tibetan text image. Background technique [0002] Text is an important carrier of human development, one of the main media for information transmission, and one of the important ways for people to record history. Tibetan is my country's first national script with international standards, and it is also one of the oldest scripts in the world. Tibetan historical documents preserve the essence of Tibetan cultural thought and are the precious wealth of human cultural thought. In order to protect this ancient and precious historical and cultural heritage and facilitate people to consult according to the content of the text, converting images of Tibetan ancient books into text is an important method to protect Tibetan historical documents. [0003] Generally speaking, the transformation of ancient book images into computer-readable text needs to go th...

AI Technical Summary

Problems solved by technology

This method has two disadvantages in dealing with Tibetan historical documents: (1) It can only estimate the approximate position of the text line in the document, and cannot deal with the curved text lines that exist in a large number of Tibetan historical documents

(2) For the cohesive parts in Tibetan historical documents, traditional projection-based segmentation methods cannot accurately locate and segment

Images