Words and image dividing method on the basis of adjacent edge point distance statistics

Abstract

The text image segmentation method based on adjacent edge point distance count in image processing field includes: detecting grey image edge with Canny operator to obtain horizontal distance between adjacent edge points; counting probable text edge point distances and detecting possible text image rows to form the text image row region; counting edge point number, detecting precise text image regions and calculating text image height; counting number of image pixel edge points, setting threshold of possible text image columns and detecting possible text image columns to form text image column region and obtain precise text image column region via column edge regulation; and detecting the positions of rows and columns to obtain the range of text image regions and segment text image finally. The present invention has raised text image segmenting precision and algorithm speed.

Description

technical field [0001] The invention relates to a character image segmentation method, in particular to a character image segmentation method based on the statistics of the distance between adjacent edge points. It belongs to the field of image processing. Background technique [0002] Through literature search, it is found that Hua Yang et al published "Extraction of Bibliography Information Basedon Image of Book Cover" ("Extracting bibliographic information from the image of the book cover"), the method proposed in this paper first converts the RGB (red, green, blue) color space of the image to HSI (color degree, saturation, intensity) color space, and then perform color segmentation and connected domain analysis to segment text from the book cover image. [0003] The technology involved in this paper mainly has the following defects and deficiencies: (1) the algorithm is limited to text image segmentation from the hardcover book cover; (2) the algorithm execution time i...

AI Technical Summary

Problems solved by technology

[0003] The technology involved in this paper mainly has the following defects and deficiencies: (1) the algorithm is limited to text image segmentation from the hardcover book cover; (2) the algorithm execution time is very long, and it takes about 320 seconds to complete the algorithm once; (3) The size of characters in the same image is different, and the stroke width of characters is also different. It is difficult to accurately detect the stroke width of characters by run-length histogram statistics; (4) Although the algorithm has been verified by several experiments, it needs to be further adopted. (5) There is no reference to the resolution of the image, or the size of the image in the above document

When solving the above problems, there are mainly the following difficulties: (1) The text size varies widely, the largest text is 60×48 pixels, and the smallest text is 10×8 pixels

(2) There are many types of text colors, and some text colors are similar to the background color, so it is difficult to effectively perform color segmentation

Images