Blur Estimation for Liveness Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user verification systems face challenges in accurately distinguishing between living and non-living objects using blur estimation, often resulting in false acceptances due to fixed focal distances and erroneous blur calculations, which can misclassify true faces as fake and vice versa.
Innovation Solution
A processor-implemented method that acquires size information of an input image, resizes it to a preset size, estimates blur based on the size ratio, and performs a liveness test by comparing the estimated blur to an adaptively varying threshold, using a neural network-based blur estimation model to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If blur estimation is performed on images of fixed size, then processing speed is improved, but measurement precision deteriorates due to size-related blur variations
Solution Approach 1:
The patent applies dynamics by making the blur threshold adaptive rather than fixed. The threshold varies according to the size of the input image, allowing the system to maintain high measurement precision across different image sizes while preserving processing efficiency. This is achieved by dynamically adjusting the threshold based on image dimensions without requiring complex recalibration.
Solution Approach 2:
The patent changes the parameter of blur threshold from a fixed value to a variable that depends on image size. By establishing a relationship between image size and appropriate blur threshold, the system maintains accurate blur estimation across varying input conditions while avoiding the computational overhead of full image processing for each size variation.
2Device complexity
If blur threshold is fixed, then device complexity is reduced, but reliability deteriorates due to false acceptances and rejections
Solution Approach 1:
The system transitions from a static blur threshold to a dynamic one that adapts to image size variations. This dynamic adjustment significantly improves verification reliability by reducing false acceptances and rejections, while maintaining relatively simple system architecture through a straightforward threshold adjustment mechanism based on image dimensions.
3Adaptability or versatility
If image size varies, then adaptability is improved, but measurement precision deteriorates due to inconsistent blur estimation
Solution Approach 1:
The patent addresses measurement consistency across varying image sizes by changing the blur threshold parameter according to image size. This allows the system to maintain high adaptability to different input image dimensions while preserving measurement precision through size-appropriate threshold selection, avoiding the need for complex normalization or resampling procedures.
Data Source
AI summary
A processor-implemented method with blur estimation includes: acquiring size information of an input image; resizing the input image to generate a target image of a preset size; estimating a blur of the target image; and estimating a blur of the input image based on the size information of the input image.