Character Separation Method of Consecutive Strings in Complicated Environment

Abstract

The invention discloses a character separation method for consecutive character strings in a complex environment, and belongs to the technical field of image processing. The invention uses the envelope information of the character string to initially locate the primary division point, and then generates the secondary division point according to the filtering process of the primary division point, and then obtains the primary division point, and finally calibrates the primary division point to accurately locate Segmentation position, so as to realize the cutting processing of the collected word string image, and achieve the purpose of separating tightly connected word strings in complex environments. The application of the present invention realizes the segmentation of tightly connected character strings in a complex environment, and is applicable to different fonts, with low computational complexity and high segmentation accuracy.

Description

technical field [0001] The invention relates to a character separation method for closely connected strings, and is especially suitable for the segmentation method of number strings in various real bills. Background technique [0002] Bills are frequently used tools in daily life and work. In order to more conveniently convert some information on the bills into digital information for subsequent processing, some automatic identification and processing methods are usually used. A very important step before recognizing characters is to split the string to be recognized. However, the complex background information and closely cohesive characters in the bill are the difficulties faced by various segmentation methods at present. Therefore, the problem of segmenting tightly connected strings in complex environments has great research value and application prospects. [0003] There are three commonly used segmentation methods at present. The first is the horizontal projection met...

AI Technical Summary

Problems solved by technology

[0003] There are three commonly used segmentation methods at present. The first is the horizontal projection method, that is, according to the horizontal direction, calculate the lowest point of the projection in the vertical direction, and obtain the segmentation position belonging to the target character. , the feasibility of this method will be greatly reduced, background noise and adhesion will affect the accuracy of projection segmentation, and thus wrongly calculate the location of the segmentation

Another method that is often used at present is to segment characters through the connectivity of characters, that is, to segment according to the principle that pixels belonging to the same character form a connected area. This method is clear and connected to character points such as license plate recognition. It is very practical when the area is obvious, but it is not applicable to the situation where characters are often glued in real bills, and the accuracy rate is very low

The third is the cutting algorithm based on template matching. According to the structure and size characteristics of the string, a string template is designed, and such a template is used to slide and match the license plate area to cut the characters to determine the best matching position. However, this method, There will be a big problem with various fonts in various bills, because it only applies to one font, or different fonts with little change

Images