Fingerprint Sensing Neural Network Extension
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fingerprint identification technologies face challenges with small area fingerprint images due to insufficient features and difficulties in image stitching, leading to potential misjudgments and failed image stitching when sensing distance is limited.
Innovation Solution
A fingerprint sensing apparatus and method that uses a neural network model to generate predicted extension blocks from image edge blocks, merging them with original fingerprint images to create extended images with more features, improving matching accuracy and success rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If the sensing distance of the fingerprint sensor is limited, then the device structure is compact, but the fingerprint image size becomes small resulting in insufficient fingerprint features
Solution Approach 1:
The patent uses a neural network model to generate a predicted extension block that copies and extends the fingerprint pattern from the image edge block. This virtual copying approach creates additional fingerprint image area without physically expanding the sensor, thereby increasing fingerprint features while maintaining the compact sensor design.
2Area of stationary object
If multiple small area fingerprint images are stitched together, then the complete fingerprint image can be obtained, but the image stitching fails when there are no similar image blocks corresponding to the same fingerprint part
Solution Approach 1:
The patent performs preliminary action by generating the predicted extension block before the image stitching process. This extension block provides additional fingerprint features that can serve as reference points for matching and stitching multiple fingerprint images, thereby improving the reliability of the image stitching process.
Solution Approach 2:
The predicted extension block acts as an intermediary that bridges the gap between multiple small fingerprint images. By providing extended fingerprint patterns from edge blocks, it creates additional overlapping regions that facilitate successful matching and stitching of multiple images.
3Area of stationary object
If directly enlarging a small area fingerprint image is performed, then the image size increases, but the amount of fingerprint information does not increase
Solution Approach 1:
The patent changes the parameter of image area by generating an extension block that increases the fingerprint image size. Unlike simple enlargement, this approach uses neural network prediction to generate meaningful fingerprint patterns in the extended area, thereby increasing both image size and the quantity of valid fingerprint information simultaneously.
Data Source
AI summary
A fingerprint sensing apparatus and a fingerprint identification method are provided. An original fingerprint image is obtained by the fingerprint sensor. An image edge block located at an edge of the original fingerprint image is selected. The image edge block is input into a neural network model to generate a predicted extension block. An extended fingerprint image is generated through merging the original fingerprint image with the predicted extension block. A fingerprint application is executed according to the extended fingerprint image.