System and method for implementing algorithmic correction of image distortion within a fingerprint imaging system

• US 20070058844A1
• Filed: 09/06/2006
• Published: 03/15/2007
• Est. Priority Date: 09/06/2005
• Status: Active Grant

• Current Assignee: Identification International, Inc.
• Original Assignee: Identification International, Inc.
• Inventors: Fenrich, Richard
• Granted Patent: US 8,068,646 B2
• Time in Patent Office: Days
• US Class Current: 382/124
• CPC Class Codes: G06T 7/80 Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
• CPC Class Codes: G06V 40/1324 by using geometrical optics, e.g. using prisms G06V40/1312 takes precedence
• CPC Class Codes: G06V 40/1347 Preprocessing; Feature extraction

Abstract

A software implemented system and method for algorithmic correction of systematic image distortions within fingerprint imaging systems. The system and method may implement a three dimensional geometric model of a fingerprint imaging system to discover where a configuration prescribed by a conceptual fingerprint imaging system and an actual configuration of a manufactured fingerprint imaging system differ. By describing this difference using the geometric model, images captured by the manufactured fingerprint imaging system can be rectified in operational use to generate rectified images with relatively low amounts of distortion present. Rectifying the images to remove distortion based on the geometric model, without physically adjusting and/or correcting the manufactured fingerprint imaging system or its components, may enable the fingerprint imaging system to be manufactured with relatively lower tolerances without degrading a precision of the images generated by the system. This may enable an enhancement in the precision of generated images without increasing a cost (e.g., time costs, monetary costs, etc.) of the fingerprint imaging system (or its components) generating the images.

Claims

1. A method of calibrating a geometric model of a fingerprint imaging system, the geometric model including a plurality of surface parameters that define the relative positions of surfaces within the fingerprint imaging system at which light is processed, the method comprising:

capturing an image of a predetermined target with the fingerprint imaging system, the target including distinguishing points provided at predetermined positions on the target;

measuring positions of the distinguishing points in captured image of the target;

comparing the measured positions of the distinguishing points in the captured image to positions of the distinguishing points predicted according to the geometric model of the fingerprint imaging system; and

adjusting the surface parameters based on the comparison between the measured positions of the distinguishing points and the predicted positions of the distinguishing points.

2. The method of claim 1, wherein comparing the measured positions of the distinguishing points to the predicted positions of the distinguishing points comprises determining evaluation functions that describe a predicted difference between measured positions and predicted positions of a set of one or more distinguishing points, the surface parameters being parameters in the evaluation functions.

3. The method of claim 1, wherein comparing the measured positions of the distinguishing points to the predicted positions of the distinguishing points comprises determining a value of a metric function for the surface parameters, wherein the metric function describes an accuracy of the geometric model in predicting the positions of the distinguishing points in the captured image of the target as a function of the surface parameters.

4. The method of claim 3, wherein the step of adjusting the surface parameters is performed as part of a curve-fitting algorithm designed to minimize the metric function.

5. The method of claim 4, wherein the curve-fitting algorithm comprises an iterative non-linear algorithm.

6. The method of claim 5, wherein the curve-fitting algorithm comprises a Levenber-Marquardt algorithm or a Gauss Newton algorithm.

7. The method of claim 1, further comprising storing the adjusted surface parameters.

8. The method of claim 7, further comprising rectifying an image captured with the fingerprint imaging system based on the adjusted and stored surface parameters.

9. A method of determining a geometric model of a fingerprint imaging system that captures images of an object located in an object plane of the system, the images being captured in an imaging plane of the system in which an image of the object is formed, the method comprising:

determining information related to the position of one or more optical surfaces within the fingerprint imaging system, wherein the one or more optical surfaces comprise surfaces within the fingerprint imaging system at which light is processed within the fingerprint imaging system;

defining the positions of the one or more optical surfaces, wherein the position of a given optical surface is defined by one or more surface parameters of the optical surface;

determining one or more evaluation functions of the fingerprint imaging system that enable the prediction of locations of visual information in the image plane of the fingerprint imaging system based on the locations of the corresponding visual information in the object plane of the fingerprint imaging system, wherein the one or more surface parameters are parameters of the one or more evaluation functions;

capturing an image in the image plane of the fingerprint imaging system, wherein the captured image is of a predetermined target located at the object plane of the fingerprint imaging system, the target including distinguishing points provided at predetermined positions thereon;

predicting positions of the distinguishing points in the captured image of the predetermined target based on the one or more evaluation functions;

measuring positions of the distinguishing points in the captured image of the predetermined target;

comparing the measured positions of the distinguishing points in the captured image to the predicted positions of the distinguishing points in the captured image predicted in accordance with the one or more evaluation functions; and

adjusting the surface parameters based on the comparison between the measured positions and the predicted positions of the distinguishing points in the captured image.

10. The method of claim 9, wherein comparing the measured positions of the distinguishing points to the predicted positions of the distinguishing points comprises determining a value of a metric function for the surface parameters, wherein the metric function describes an accuracy of the geometric model in predicting the positions of the distinguishing points in the captured image of the target as a function of the surface parameters.

11. The method of claim 10, wherein the step of adjusting the surface parameters is performed as part of a curve-fitting algorithm designed to minimize the metric function.

12. The method of claim 11, wherein the curve-fitting algorithm comprises an iterative non-linear algorithm.

13. The method of claim 12, wherein the curve-fitting algorithm comprises a Levenber-Marquardt algorithm or a Gauss Newton algorithm.

14. The method of claim 9, wherein the measured positions of the distinguishing points and the predicted positions of the distinguishing points comprise positions of the distinguishing points in relation to each other.

15. The method of claim 9, further comprising storing the adjusted surface parameters for rectifying images captured by the fingerprint imaging system.

17. The method of claim 11, wherein the curve-fitting algorithm comprises an iterative non-linear algorithm.

18. The method of claim 12, wherein the curve-fitting algorithm comprises a Levenber-Marquardt algorithm or a Gauss Newton algorithm.

16. A method of calibrating a geometric model of a fingerprint imaging system, the geometric model including a plurality of surface parameters that define the relative positions of surfaces within the fingerprint imaging system at which light is processed, the method comprising:

capturing an image of a predetermined target with the fingerprint imaging system, the target including distinguishing points provided at predetermine positions;

predicting the positions of one or more of the distinguishing points in the captured image of the predetermined target based on the geometric model;

determining a value of a metric function that describes an accuracy of the geometric model in predicting the positions of the distinguishing points in the captured image of the target as a function of the surface parameters included in the geometric model; and

implementing a curve-fitting algorithm that adjusts the surface parameters of the geometric model to reduce the value of the metric function.