Have you ever thought about why the process of verifying identity works so fast nowadays? Traditionally, Know Your Customer (KYC), which is a procedure of identifying and verifying customers usually used by financial companies or institutions, was executed manually by a human personnel. At that time, The verifying process was far from simple and could take minutes if not hours. You need to go to the office in person, queue, wait for a staff member to inspect the documents, answer their questions, let them compare your answers and real life appearance with the documents, etc.
However, with AI technologies and their capabilities in understanding language and image processing, verifying a person’s identity by their ID card or document is no longer a hassle! One of the most influential technologies that can help us is called Optical Character Recognition (OCR).
Optical Character Recognition (OCR) is one of many AI systems that can automate data extraction from scanned documents, photographs of texts, and image-only PDFs. Machines can’t comprehend a photo like how humans see shapes and read text from pictures. However, with OCR technology, the system can recognize text in images and convert them into machine-readable digital files which can also be edited and analyzed by other softwares.
Understanding the mechanisms behind OCR can help us to better comprehend its precise contributions to the e-KYC process. In general, the operational framework of OCR involves four key steps:
In the first stage, OCR scans the image and converts the document into binary data. This step is crucial as it distinguishes various brightness levels within the image, classifying light areas as background and dark areas as text, or vice versa.
The OCR software also takes charge of improving the quality of acquired images by utilizing a series of its image processing capabilities. OCR can refine images by correcting skewed images, eliminating noise through despeckling, and cleaning up boxes and lines. In addition, an OCR program can also analyze the structure of a document, dividing the page into elements such as blocks of texts, tables, or images.
After identifying the text areas, the third and most crucial step is the text recognition. This stage involves the use of two primary OCR algorithms: pattern matching and feature extraction. Pattern matching compares character images (glyphs) with stored examples of similar font and scale. This is a method particularly suitable for known font documents. On the other hand, feature extraction breaks down glyphs into distinctive features like angled lines, crossed lines, or curves to identify the best match. This approach offers adaptability when faced with specific letters or numbers with unique features.
Finally, following successful text recognition, the system converts the recognized text data into a coherent digital file. This last step ensures that the extracted information can be easily utilized for further analysis or integration into various digital systems.
The integration of OCR technology in e-KYC mainly focus on their text and image processing abilities:
OCR can automate the extraction of crucial text information from ID cards and documents, such as name, date of birth, home address, occupation, etc. This automation accelerates the process of cross checking data and minimizes the likelihood of manual errors that may arise during typing or transcribing.
As previously stated, OCR has the ability to process images up to a certain extent, with auto cropping falling within its capabilities. Through the automated cropping of photos and signatures from ID cards, OCR plays a crucial role in streamlining face matching and signature verification processes, ultimately saving time and improving precision.
Another advantage stemming from OCR and its image processing capabilities is its ability to conduct a thorough evaluation of the image quality. This includes the assessment of various quality parameters such as:
By ensuring that submitted images meet the required standards, OCR significantly enhances the overall reliability and accuracy of the overall e-KYC process.
All in all, the integration of AI into the e-KYC process is predominantly driven by OCR technology, specifically in automating text extraction and image cropping. Additionally, OCR also enables a quick evaluation of the quality of the photo submitted by the user. The capabilities of OCR have introduced a new, faster, and more accurate approach to the overall identity verification process, particularly for photos and signatures in ID card verification.
