Optical Character Recognition (OCR) is the identification of scribed, stamped, printed or pre-printed text. This can be carried out in all languages, fonts, sizes and styles. Optical Character Verification (OCV) is the authentication that an identified character string matches expected values stored in your PLC, database or FTP server.
Machine vision OCR is used to automatically read printed, scribed or stamped text at high speed to confirm text readability, quality and form. More detailed analysis such as missing or touching characters, dimensional checks, skew and distortion can also be analysed. In addition to this, OCV tools can also be used to verify text and numbers against expected results communicated from a PLC, printer, database or FTP server. IVS latest generation deep OCR uses Ai to read even under the most difficult conditions such as distorted letters and inconsistent surfaces with background noise.
Extremely fast and powerful character recognition algorithms can verify text strings in fractions of a second, ensuring your production isn’t compromised.
Subconsciously, the human brain may read incorrectly spelt words and mistakenly decipher the information to understand them correctly. Machine vision eliminates the human error factor and consistently produces accurate results.
Keep track of all read text with spreadsheet or SQL data logging for comprehensive production traceability.
The OCR wizard within our vision systems allows new datasets and styles to be added easily to enable the training, classification and verification of character sets.
Powerful algorithms locate text and easily identifies characters that are translated, rotated and scaled to sub-pixel accuracy. IVS® neural network tools can learn to recognise characters, symbols and digits, etc.
Once the vision system has identified a potential character candidate, it is assessed against the trained character set. If the candidate is a match, the vision system will identify the candidate as a readable character and output the read result.
There are a variety of ways an IVS vision system can recognise and classify text, these include:
Matching and OCR Tools: pre-trained ready to use fonts are available in the toolkit or each character can be trained individually to allow for greater control for complex symbols etc. The system will quickly locate and read all characters in the region of interest.
Classifiers: this function sorts characters into classes based on their characteristics.
Ai deep learning OCR: The latest IVS® machine vision technologies mimic human brain activity when learning a task, allowing vision systems to identify and read text in the same way that humans can
We have a library of pre-trained fonts, characters, digits and symbols from a vast amount of application areas that can be utilised to ensure your OCR applications are executed quickly and accurately. We understand each application is unique, therefore our training wizard also allows further optimisation for each application.
Our vision systems can read in any language, whether printed, laser engraved, dot printed, and even handwritten text. Pre-trained fonts for OCR-A and OCR-B are also included in our systems.
Once the text has been read, each character in the string is evaluated for quality and class. A 100% quality score would indicate a perfect match for any given character. The class verification is simply analysing if the candidate is the correct character.
The read text can be sent to a factory automation system, appended to filenames for image archiving, and kept in a spreadsheet or SQL database, allowing the part to be tracked completely.
OCR-A and OCR-B fonts were designed during the early progression of OCR to aid the recognition process of printed characters. All characters have the same thickness and width and are distinctly different from each other.
With the latest machine vision technology, the need to use OCR fonts has been reduced and even handwritten characters can now be read by machine vision systems. Some manufactures still choose to use OCR fonts because they are easy to read and each character is unique.