Fingerprint Sensor Image Compensation for Flexible Displays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Obtaining satisfactory fingerprint image data from fingerprint sensors deployed in flexible display devices is challenging due to issues like background patterns caused by flexible backer layers and varying forces applied during fingerprint scanning.
Innovation Solution
A method involving a control system that receives fingerprint image data, obtains background image data, processes the data to filter out background patterns using force data and machine learning models, and outputs enhanced fingerprint image data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fingerprint sensor is deployed in a flexible display device, then the device can achieve flexible form factor and adaptability, but background patterns from the flexible backer layer degrade fingerprint image quality
Solution Approach 1:
The patent extracts and removes the harmful background patterns from the flexible backer layer from the fingerprint image data through image processing techniques. The system separates the fingerprint features from the backer layer artifacts, effectively taking out the disturbing background elements while preserving the fingerprint information.
Solution Approach 2:
The patent changes parameters such as force compensation and uses machine learning models to adapt the image processing based on detected force levels. By adjusting processing parameters dynamically based on force data, the system optimizes fingerprint image quality while accounting for the flexible backer layer's influence under different pressing conditions.
2Productivity
If force is applied during fingerprint scanning, then fingerprint image data can be obtained, but varying forces cause background pattern variations that reduce scanning accuracy
Solution Approach 1:
The patent implements force feedback by detecting the applied force during fingerprint scanning and using this information to compensate for background pattern variations. The system measures the force applied and adjusts the image processing accordingly, creating a closed-loop system that maintains scanning accuracy across different force levels.
Solution Approach 2:
The patent makes the image processing dynamic by adapting to varying force conditions in real-time. Instead of using a fixed processing approach, the system dynamically adjusts background subtraction and enhancement parameters based on the detected force, allowing accurate fingerprint capture under diverse pressing conditions.
3Measurement precision
If machine learning models are used to process fingerprint image data, then image quality and accuracy are improved, but power consumption and computational resources increase
Solution Approach 1:
The patent applies partial machine learning processing by using force data to determine when and how to apply complex ML models. Instead of always using full ML processing, the system selectively applies ML-based background pattern removal based on force conditions, reducing unnecessary computational overhead while maintaining accuracy when needed.
Data Source
AI summary
Some methods may involve receiving fingerprint image data and a first set of background image data from a fingerprint sensor and determining first processed fingerprint image data via a subtraction of the first set of background image data from the fingerprint image data. Some methods may involve obtaining force data corresponding to a force applied to the fingerprint sensor when the fingerprint image data were obtained. Some methods may involve obtaining a second set of background image data corresponding to the force data. Some methods may involve determining second processed fingerprint image data based, at least in part, on the first processed fingerprint image data and the second set of background image data, and outputting the second processed fingerprint image data. In some examples, determining the second processed fingerprint image data may involve a machine learning model. Some examples may involve estimating residual noise based on the force data.