Method and device for recognizing nixie tube text and electronic equipment

Abstract

The invention relates to a method and device for recognizing a nixie tube text and electronic equipment, and belongs to the technical field of character recognition, and the method for recognizing the nixie tube text comprises the steps that a field collection image of target equipment is obtained; selecting an equipment model graph matched with the target equipment from a pre-constructed model graph library, and recognizing and deducting the field acquisition image according to the equipment model graph to obtain a to-be-recognized area image; based on the to-be-recognized area image, performing text recognition by adopting a pre-constructed and trained nixie tube text recognition model to obtain a text recognition result; and performing matching combination processing according to text attribute configuration information associated with the equipment model diagram and the text recognition result, and taking structured data obtained by combination as a final recognition result. According to the invention, recognition of nixie tube display is realized better.

Description

technical field [0001] The application belongs to the technical field of character recognition, and in particular relates to a method, device and electronic equipment for recognizing digital tube text. Background technique [0002] Optical Character Recognition (OCR) generally refers to the process of using electronic equipment (scanners, digital cameras, etc.) to check the text printed on paper, by detecting light and shade and shape, and using character recognition technology to translate it into text. Tesseract-OCR is widely used in traditional OCR recognition applications. This software was developed by Hewlett-Packard's Bristol laboratory in 1984-1994 and was originally used as a text recognition engine for Hewlett-Packard's flatbed scanners. In 1995, it won the first place in the UNLV OCR character recognition accuracy test and received widespread attention. After 1994, the development was stopped because HP gave up the OCR market. In 2005, HP contributed Tesseract-O...

AI Technical Summary

Problems solved by technology

[0005] As mentioned above, traditional and complex scene OCR has had good solutions and implementations in practical applications, but in specific scenarios, in the existing related technologies, text displayed on digital tubes on the device display screen (including displaying text in digital tube fonts) The recognition effect of the text) is very unsatisfactory

[0006] Specifically, traditional OCR is mainly aimed at text recognition on printing paper. For simple scenes, the difference between text and background is large, and the binarization scene has a significant effect; in actual application scenarios, there are many interference factors, and the original image includes the entire device and operating environment. Affected by various factors such as light and shade, angle, color, etc., irrelevant information is often recognized as useful information, and text objects cannot be effectively extracted, and the existing model does not support seven-segment digital tube text recognition

[0007] The OCR recognition function of complex scenes supports seven-segment digital tube target positioning, but there is no algorithm model support. It takes a lot of time to design, train, and optimize the network. The training requires a large number of full-scale equipment types, and effectively annotated pictures of scenes with different light and shade, angles, and colors. , this part of the work requires a lot of manpower and material resources, which cannot be provided on site. The trained network needs to be continuously verified and optimized. When new types of devices are added, the network needs to be retrained and verified and optimized. When rapid development is required, this part of the workload cannot estimate

At the same time, it also has the defect that the existing model does not support text recognition of seven-segment digital tubes, so it supports OCR functions in complex scenes, but it requires a lot of work from design implementation to optimization, and the recognition effect cannot be guaranteed, and the recognition result is a piece of text, there is no Any data attribute business information

Images