Adhesion character recognition method and system

Abstract

The invention relates to an adhesion character recognition method and system. The method comprises the following steps: detecting and intercepting a target area; denoising and binarization processingare carried out on the obtained target area cutting image; performing row segmentation on the two rows of character binary images by using a horizontal projection method; performing column segmentation on the single-row character image based on the character width and a vertical projection method; and recognizing the characters based on the convolutional neural network. The invention provides a segmentation algorithm based on character width, utilizes pixel difference between adjacent characters and combines a projection method to realize segmentation of adhered characters, and also provides an adhered character recognition system which can perform online recognition on the adhered characters and irregular characters, and is good in adaptability and high in practicability. The recognitionrate of adhered characters and irregular characters is greatly improved.

Description

technical field [0001] This patent application belongs to the technical field of inkjet character recognition, and more specifically relates to a method and system for recognizing glued characters. Background technique [0002] In the modern age of rapid development of informatization, character recognition is an important method of entering and converting information. It has a very wide range of applications in the fields of license plate recognition, postal code recognition, and dairy product coding recognition. With the improvement of living standards, milk and dairy products have become an indispensable part of people's lives. Therefore, the quality monitoring of dairy production becomes particularly important. [0003] The current inkjet character recognition partly relies on manual detection. The disadvantages of this detection are high cost and slow detection speed, which cannot meet the requirements of industrialization. The other part is to rely on the convention...

AI Technical Summary

Problems solved by technology

However, in the process of coding, due to the instability of machine operation and other factors, it is inevitable that the characters of the coding will be incomplete, the characters will be confused and sticky, etc.

Using traditional template matching and other methods for character recognition, for irregular characters or sticky characters, the adaptability is poor, the recognition rate is low, and the recognition method needs to be further improved

Images