Onfido has filed a patent for computerized methods and systems to authenticate physical documents. The system uses a mobile device with an image capture device to capture a sequence of images of the document while tracking it and adjusting imaging parameters. The device then selects images, classifies the document, identifies a region of interest, reconstructs it, generates an authentication score, and determines the document’s authenticity. GlobalData’s report on Onfido gives a 360-degree view of the company including its patenting strategy. Buy the report here.
According to GlobalData’s company profile on Onfido, Liveness detection was a key innovation area identified from patents. Onfido's grant share as of September 2023 was 28%. Grant share is based on the ratio of number of grants to total number of patents.
The patent filed is for a system to authenticate physical documents

A recently filed patent (Publication Number: US20230281820A1) describes a system for authenticating physical documents using a mobile computing device and an image capture device. The system captures a sequence of images of the document as it is rotated, tracking the document throughout the sequence. It adjusts the operational parameters of the image capture device based on imaging conditions associated with the document. The system then selects and classifies images from the sequence, identifies a region of interest in the document, and reconstructs that region using the selected images. An authentication score is generated based on the reconstructed region, and the system determines whether the document is authentic based on this score.
The system allows for rotation or tilting of either the document or the image capture device along multiple axes. It dynamically determines a minimum range of motion for the document and instructs the user to continue rotating or tilting until this range is satisfied. The system also dynamically adjusts lighting parameters of the image capture device and assesses the signal associated with the region of interest to ensure a minimum amount of signal is captured.
The system tracks the document's location or six-dimensional pose in each image of the sequence and adjusts operational parameters such as shutter speed, ISO speed, gain, aperture, flash intensity, flash duration, or light balance based on imaging conditions. It selects usable images for authentication by discarding unusable images that may be occluded, have excessive viewing angles, contain excessive noise, or are blurry.
The system identifies the region of interest in the document by detecting its location and estimating its pose in each selected image. It then crops and aligns the images based on the estimated characteristics of the document. The region of interest may include an optical variable device (OVD).
To generate an authentication score, the system uses machine learning classification models such as deep learning models, Random Forest algorithms, Support Vector Machines, neural networks, or ensembles thereof. The classification value generated by these models represents the probability, confidence score, or similarity metric indicating the authenticity of the document. Interpretable methods, such as occlusion, perturbation, or heatmap analysis, can be used to validate the classification value.
The system can be trained using a dataset of genuine and fraudulent documents, and the classification value can measure the similarity between these documents. Prior to selecting images, the system preprocesses the sequence by assessing video quality metrics, detecting
To know more about GlobalData’s detailed insights on Onfido, buy the report here.
Data Insights
From
The gold standard of business intelligence.
Blending expert knowledge with cutting-edge technology, GlobalData’s unrivalled proprietary data will enable you to decode what’s happening in your market. You can make better informed decisions and gain a future-proof advantage over your competitors.