ID Document Image Rotation and Alignment for eKYC Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing technologies face reduced accuracy when handling misoriented ID document images, as they are not effective for images with bottom edges misaligned with the x-axis, leading to inefficiencies in eKYC processes.
Innovation Solution
An image processing method and system that involves determining and correcting the rotation of ID document images to align the bottom edge with the x-axis, followed by image alignment and regression, allowing for the use of smaller models for increased processing speed and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current image processing technologies are applied to misoriented ID document images, then processing can be performed, but accuracy is reduced
Solution Approach 1:
The patent applies preliminary rotation to the ID document image before performing image alignment. The system determines the rotation angle of the ID document image and rotates it to a predetermined angle range, ensuring the image is properly oriented before subsequent processing steps. This preliminary action resolves the contradiction by preparing the image in advance, allowing accurate processing even when the original image is misoriented.
2Measurement precision
If larger models are used to improve accuracy of image alignment on misoriented images, then alignment accuracy improves, but processing speed decreases
Solution Approach 1:
By rotating the ID document image to a predetermined angle range before alignment processing, the system transforms misoriented images into properly oriented ones. This allows the use of smaller, faster models for image alignment while maintaining high accuracy, as the models only need to handle images within a limited angle range rather than all possible orientations.
Solution Approach 2:
The patent changes the rotation angle parameter of the ID document image to fall within a predetermined range (e.g., -30 to 30 degrees). This parameter transformation simplifies the subsequent alignment task, enabling the use of more efficient models that operate faster while achieving the same or better alignment accuracy on pre-oriented images.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image processing method and system are provided. The method includes determining a rotated angle of an identification (ID) document image, wherein the rotated angle corresponds to an angle between a bottom edge of the ID document image and an x-axis; rotating the ID document image based on the determined rotated angle, such that an angle between the bottom edge of the rotated ID document image and the x-axis is within a predefined range; performing image regression on the rotated ID document image to obtain coordinates of four corner points of the rotated ID document image; and performing image alignment on the rotated ID document image based on the obtained coordinates of four corner points of the rotated ID document image.