Fingerprint Image Segmentation and Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fingerprint capture systems are not adaptable to different image sensors, as methods suited for optical or TFT sensors cannot function correctly with images from the other type, leading to inconsistent image quality and stability issues due to varying image acquisition frequencies.
Innovation Solution
A method that involves image segmentation, region tracking, and confidence mark adjustment based on image acquisition frequency to generate a final fingerprint image, allowing for adaptation to any type of image sensor by determining admissible regions and ending image acquisition when a predetermined threshold is reached, ensuring efficient image processing regardless of sensor type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If image segmentation and region tracking methods are used with fixed confidence thresholds, then processing is simple, but the method cannot adapt to different image sensor types with varying acquisition frequencies
Solution Approach 1:
The patent applies dynamics by making the confidence mark increment dynamic rather than fixed. The increment value is adjusted based on the image acquisition frequency of the specific sensor type, allowing the same processing method to adapt to different sensors (optical, TFT, etc.) with varying acquisition frequencies without requiring completely different processing algorithms for each sensor type
Solution Approach 2:
The patent changes the parameter of confidence mark increment based on image acquisition frequency. By modifying this key parameter according to the sensor's acquisition frequency, the system achieves adaptability across different sensor types while maintaining the same overall processing framework, thus improving versatility without proportionally increasing complexity
2Productivity
If image acquisition frequency is increased to reduce user wait time, then productivity improves, but image quality may deteriorate due to shorter acquisition time
Solution Approach 1:
The patent applies feedback by continuously updating the confidence mark for each region based on its persistence and consistency across multiple images in the sequence. This feedback mechanism allows the system to quality-assess each potential fingerprint region dynamically, ensuring that only regions with sufficient quality and consistency contribute to the final image, thereby maintaining image quality even when acquisition frequency varies
Solution Approach 2:
The patent processes multiple images in a sequence rather than relying on a single image. By accumulating evidence across multiple partial observations (individual images), the system builds a more complete and reliable final fingerprint image. This approach allows the system to tolerate variations in individual image quality while maintaining overall image quality through the cumulative effect of processing several images
3Measurement precision
If a sequence of images is captured to improve image quality, then measurement precision improves, but loss of time increases due to multiple acquisitions
Solution Approach 1:
The patent applies preliminary action by pre-defining confidence thresholds and increment values based on expected image acquisition frequencies. This allows the system to quickly process image sequences without requiring complex real-time adjustments, reducing computational overhead and processing time while still achieving high measurement precision through the sequence processing
Solution Approach 2:
The patent adjusts the confidence mark increment parameter based on the image acquisition frequency. When acquisition frequency is high, the increment is adjusted accordingly, allowing the system to reach the confidence threshold faster and terminate the sequence earlier. This dynamic parameter adjustment optimizes the balance between image quality and acquisition time based on the specific sensor being used
Data Source
AI summary
A method for obtaining an image of at least one fingerprint from a sequence of images is disclosed. The method comprises a segmentation of images in a sequence of acquired images making it possible to determine regions liable to comprise a fingerprint, a tracking of regions determined along with the sequence of images and a calculation of a confidence mark for each region determined. The confidence mark of a region determined is incremented (or respectively decremented) at each acquired image if the region is present in (or respectively absent from) the image, the increment (or respectively the decrement) being dependent on the predefined image acquisition frequency. Information making it possible to obtain a final image containing fingerprints is generated by the method when a predefined number of regions each having reached a confidence mark at least equal to a predetermined threshold is obtained.