Method for positioning and identifying tire embossed characters

Abstract

The invention discloses a method for positioning and identifying tire embossed characters. The technical problem mainly solved by the invention is to position and identify target embossed characters on an automatic assembly line of tires and hubs. The method comprises the following steps: acquiring tire images in a darkroom by using an industrial camera; preprocessing the collected images; training the positioning module; training the identification module; and verifying the identification result through a character string rule. The method for positioning and identifying tire embossed characters mainly aims at a certain type of specific tire, can accurately position and identify the position of the target imprinting character in a short time under the condition of less data sets, has relatively high innovativeness and practicability, and lays a foundation for subsequent extraction of key information in the target character.

Description

technical field [0001] The invention relates to a character positioning and recognition method, in particular to a tire embossed character positioning and recognition method, which belongs to the technical field of machine vision and machine learning. Background technique [0002] On the automatic assembly line of tires and hubs, sometimes errors occur when loading tires and hubs. Now, machine vision is used to identify the key strings in the tires and hubs before assembly, and extract the important information. It is used to detect whether there is a mistake; the traditional method can only be observed by the naked eye, which is not only slow in efficiency, but also reduces the recognition accuracy compared with machine vision. Therefore, the combination of machine vision and deep learning It will become an inevitable trend in future automated production lines, assembly lines and testing lines. [0003] As the name suggests, the tire embossed characters are made by embossi...

AI Technical Summary

Problems solved by technology

(2) Characters are prone to incompleteness

(3) The number of molds is limited, so the fonts of embossed characters are limited and neat

(4) It is easy to appear the phenomenon of character inclination

[0004] To locate and recognize embossed characters with the above characteristics, traditional OCR technology will cause a large number of misrecognition phenomena, and on the automatic tire assembly line, its positioning and recognition environment also has the following characteristics: (1) Due to the fixed focal length of the camera field of view, At the same time, the camera field of view is required to include the entire tire, so the camera pixel is required to be very high, and the image processing time will increase accordingly

(2) Due to the assembly of tires, there may be interference factors such as stains on the characters

(3) In the context of huge pixels and complex background, the number of pixels occupied by the target string is small, and the traditional positioning method cannot directly find the position of the string

Images