OCR Character Detection Chapter| Reduce labor costs and improve production efficiency

2026-03-28 Click： 1

share

• 
  Facebook
• 
  Linkedin
• 
  Tumblr
• 
  VK
• 
  Pinterest
• 
  Copy link

Optical Character Recognition (Optical Character Recognition) refers to the process in which electronic devices (such as scanners or digital cameras) inspect characters printed on paper, determine their shapes by detecting dark and light patterns, and then use character recognition methods to translate the shapes into computer text.

The main indicators to measure the performance of an OCR system include: rejection rate, false recognition rate, recognition speed, friendliness of the user interface, product stability, ease of use and feasibility.

In the industrial world, optical character recognition (OCR) is a machine vision task that involves extracting textual information from images.

Get text location

The position of the text is fixed. For example, personal ID cards are made according to official specifications, the location of each data field is known, and a well-calibrated vision system can capture images with almost constant text location.

The position of the text is not fixed, but it is related to feature elements or special marks (optical marks) on the input image. To obtain the position of the text, the optical marks must be found. This can be done through template matching or other techniques.

Extract text from background

The main complication in the text extraction process may be uneven light. Certain techniques, such as light normalization or edge sharpening, help find characters.

segmented text

Text area segmentation is the process of splitting an area into lines and individual characters. When text lines are separated, each line must be split into individual characters, and then the extracted characters will be converted from a graphical representation to a text representation.

Call the OCR model library

By calling the OCR model library, the recognized characters are compared with the model library in text form to match the template with the most similar data to obtain accurate character information.

character recognition

Generally speaking, it is necessary to select the appropriate character specification size to classify the character size.

Application of OCR recognition and detection technology

In life, products related to printed characters can be seen everywhere, such as characters on keyboards, label characters on clothing, characters on electrical appliances, etc.

The appearance of these products will cause complaints from customers to the manufacturer due to some flaws caused by the characters during printing, such as character drawing, shifting, excessive ink, missing, leaking and other poor printing of characters.

In order to overcome this problem, manufacturers will conduct strict testing of products before leaving the factory. The traditional detection method uses manual visual inspection. Although the visual inspection method is flexible, it can make different judgments on various errors. However, due to the large quantity, variety and labor consumption of products, and long-term testing will cause eye fatigue or be confused and misjudged due to subjective emotions, which will also lead to the outflow of defective products and cause customer complaints.

Using machine vision inspection technology, it can distinguish the product defect image from the background image according to different product materials, defect conditions and customer needs, using different light sources, different illumination angles, and cameras with different pixels. Then use the color, gray scale, shape, size, etc. of the defect image to identify defects, use different algorithms to write software through picture effects, and develop software based on customer needs.

High-precision, high-efficiency, and high-stability real-time detection, analysis, and calculation of products to determine whether the product is qualified can effectively improve the detection speed and accuracy of the production line, greatly improve output and quality, reduce labor costs, and prevent misjudgment caused by human eye fatigue.

Functional characteristics of machine vision inspection technology applied to character recognition detection:

1. Inspect barcode/character printing defects, shift, excessive ink, leaks, integrity, etc.;

3. Check whether the direction of the object is correct;

4. Static or dynamic detection;

5. The OK/NG product system outputs corresponding control signals.

Hunan Yima Intelligent Technology focuses on the traceability system and online testing of one object and one code. Leading technologies cover code coding, labeling, code reading, vision, traceability, weighing, metal testing, etc.! With "tracing the origin, anti-counterfeiting and anti-theft, intelligent testing, and the Internet of Everything" as the strategic development direction, and the goal is to build a benchmark enterprise in the industrial Internet of Things industry dominated by "one object, one code".

Through years of industry practice and continuous technological innovation to improve its product research and development capabilities, we have now delivered self-developed OCR visual inspection all-in-one machines to customers in different industries many times. Vision replaces eyes and intelligence replaces manual labor. The new generation of OCR character visual inspection system can be integrated into various production lines and packaging machines, making it easy to build, connect, and use. It can start working with simple settings, becoming a dedicated "sentry" on your production line.

practical case

④ Soft package OCR recognition

Flexible packaging is the main form of food packaging, which can effectively prevent food from being damaged by external factors. Not only that, flexible packaging also needs to print important information such as production date and shelf life, which is directly related to food sales.

In order to solve the problems existing in the flexible packaging production process, Hunan Yima Intelligent Technology Co., Ltd. proposed an OCR recognition solution for the flexible packaging surface based on deep learning.

Detection difficulties

① The plastic film is transparent and is easily interfered by background patterns

② Different colors and font thicknesses

③ Character angle changes

We offer three independent product options and offer customization options to meet different requirements and budgets.

The EM-CK550 is a complete code assignment inspection and verification solution that can verify data of man-machine readable codes while checking readability and code assignment location.

EM-CK350 can check the readability of machine-readable codes;

The EM-CK150 is an affordable entry option that provides a simple check of the presence and location of assigned codes.

Automation has greatly reduced labor costs, improved production line production efficiency and character recognition accuracy, and helped high-quality testing in the food industry.

Advantages of OCR vision inspection machine:

1) Identify the wrinkled surface. Deep learning testing can accurately identify the third period, origin, lot number and other information on the package, and has no requirements on the shape of the package itself, and the text on the wrinkled part of the surface can also be accurately identified;

2) Avoid reflective interference. During recognition, the text and the background can be separated, which perfectly solves the problems of visual fatigue and difficulty in recognition due to reflection on the packaging surface and diverse colors during manual and traditional visual inspection;

3) Higher accuracy. The deep learning system only needs to perform simple manual annotation training to automatically learn defect types without the participation of professionals. As the number of samples increases, the recognition accuracy rate can approach 100% infinitely;

4) Release labor. Deep learning visual inspection can liberate workers from boring work and free up more labor.

③ Detection difficulties

1. The specifications, colors, and degree of old and new of the drug boxes vary greatly and are not uniform;

2. The size and depth of the steel stamps are different;

3. The steel seal overlaps with the text, and even has interference patterns;