System and method for normalized focal length profiling

Patent metadata

Field	Value
Publication number	US-10210602-B2
Application number	US-201615203889-A
Country	US
Kind code	B2
Filing date	Jul 7, 2016
Priority date	Oct 17, 2011
Publication date	Feb 19, 2019
Grant date	Feb 19, 2019

Abstract

A system and method are provided for normalizing a camera focal length. A perspective geometry estimator accepts a scanned image from a camera having an undefined focal length, and generates a normalized focal length estimate for the image. The normalized focal length estimate is compared to a normalized focal length density distribution. If the normalized focal length estimate meets a minimum threshold of probability, the normalized focal length estimate is selected and the image is processed using the selected normalized focal length estimate. If the normalized focal length estimate fails to meet the minimum threshold of probability, the image is processed using the highest probability prior normalized focal length from the normalized focal length density distribution.

First claim

I claim:

1. A method for normalizing a camera focal length, the method comprising:
providing an image enhancement device having a processor, a non-transitory memory, a normalized focal length density distribution database stored in the memory, and a perspective geometry estimator application stored in the memory comprising a sequence of processor executable steps for determining a normalized focal length, where a normalized focal length is a camera lens focal length determined independent of image resolution;
the perspective geometry estimator application accepting a scanned image from a camera having an undefined focal length;
generating a normalized focal length estimate for the image;
comparing the normalized focal length estimate to the normalized focal length density distribution database, wherein comparing the normalized focal length estimate to the normalized focal length density distribution database includes using a Bayes formula to compare the normalized focal length estimate to a maximum probability normalized focal length from the normalized focal length density distribution database;
selecting the generated and compared normalized focal length estimate when the normalized focal length estimate meets a minimum threshold of probability; and,
correcting the image using the selected normalized focal length estimate.

2. The method of claim 1 further comprising:
adding the selected normalized focal length estimate to the normalized focal length density distribution database.

3. The method of claim 1 further comprising:
when the selected, normalized focal length estimate fails to meet the minimum threshold of probability, discarding the normalized focal length estimate; and,
correcting the image by selecting and using a highest probability prior normalized focal length from the normalized focal length density distribution database.

4. The method of claim 1 wherein generating the normalized focal length estimate for the image includes generating a plurality of normalized focal length estimates; and,
wherein selecting the normalized focal length estimate includes selecting a maximum probability normalized focal length estimate.

5. The method of claim 1 wherein the normalized focal length density distribution database is stored in a database segregated in sections selected from a group consisting of camera manufacturer, camera model, camera serial number, and user identity.

6. The method of claim 5 wherein providing the normalized focal length density distribution database includes initially providing a normalized focal length density distribution database pre-loaded with information selected from a group consisting of camera manufacturer, camera model, camera serial number, and a history associated with a user identity.

7. The method of claim 1 wherein accepting the scanned image includes accepting a scanned image from a camera having a zoom lens;
wherein comparing the normalized focal length estimate to the normalized focal length density distribution database includes comparing the normalized focal length estimate to each one of a plurality of density distribution peaks in the normalized focal length density distribution database; and,
wherein selecting the normalized focal length estimate includes selecting the normalized focal length estimate meeting the minimum threshold of probability associated with one of the density distribution peaks.

8. The method of claim 7 further comprising:
when the selected normalized focal length estimate fails to meet the minimum threshold of probability, discarding the selected normalized focal length estimate; and,
correcting the image by selecting and using a highest probability prior normalized focal length estimate from the normalized focal length density distribution database.

9. The method of claim 7 wherein generating the normalized focal length estimate for the image includes generating a plurality of normalized focal length estimates; and,
wherein selecting the normalized focal length estimate includes selecting a maximum probability normalized focal length estimate from the plurality of normalized focal length estimates.

10. The method of claim 1 wherein generating the normalized focal length estimate for the image includes:
estimating horizontal vanishing information for the image;
estimating vertical vanishing information for the image;
performing corner detection on the image;
performing quadrilateral-boundary detection using a result of the corner detection, the horizontal vanishing information, and the vertical vanishing information, thereby identifying a first bounding quadrilateral;
determining a first plurality of geometric rectification quality measure values associated with the first bounding quadrilateral;
selecting a rectification hypothesis based on the first plurality of geometric rectification quality measure values; and,
using the selected rectification hypothesis to generate the normalized focal length estimate.

11. An image enhancement device for normalizing a camera focal length and correcting an image using the normalized camera focal length, the image enhancement device comprising:
a processor;
a non-transitory memory;
an input/output (TO) port to receive scanned camera images and to supply images;
a communications bus connecting the IO port to the processor and memory;
a normalized focal length density distribution database stored in the memory; and,
a perspective geometry estimator application stored in the memory comprising a sequence of processor executable steps for determining a normalized focal length, where a normalized focal length is a camera lens focal length determined independent of image resolution, the perspective geometry estimator application accepting a scanned image having an undefined focal length, generating a normalized focal length estimate for the image, comparing the normalized focal length estimate to the normalized focal length density distribution database, wherein the perspective geometry estimator application uses a Bayes formula to compare the normalized focal length estimate to a maximum probability normalized focal length from the normalized focal length density distribution database,
selecting the generated and compared normalized focal length estimate when the normalized focal length estimate meets a minimum threshold of probability, and supplying the image corrected using the selected normalized focal length estimate.

12. The image enhancement device of claim 11 wherein the perspective geometry estimator application adds the selected normalized focal length estimate to the normalized focal length density distribution database.

13. The image enhancement device of claim 11 wherein the perspective geometry estimator application discards the normalized focal length estimate when the normalized focal length estimate fails to meet the minimum threshold of probability, and corrects the image by selecting and using a highest probability prior normalized focal length from the normalized focal length density distribution database.

14. The image enhancement device of claim 11
wherein the perspective geometry estimator application generating the normalized focal length estimate for the image includes generating a plurality of normalized focal length estimates, and
wherein selecting the normalized focal length estimate includes selecting a maximum probability normalized focal length estimate from the plurality of normalized focal length estimates.

15. The image enhancement device of claim 11 wherein the normalized focal length density distribution database is stored in a

Assignees

• Sharp Laboratories Of America Sla Inc
• Sharp Laboratories America Inc

Inventors

• Campbell Richard John

Classifications

• G06V10/247

  by affine transforms, e.g. correction due to perspective effects; Quadrilaterals, e.g. trapezoids · CPC title

• G06K2009/363

  Physics · mapped topic

• G06T2207/30176

  Document · CPC title

• G06T2207/20076

  Probabilistic image processing · CPC title

• G06T2207/20081

  Training; Learning · CPC title