Image processing apparatus and control method therefor

Patent metadata

Field	Value
Publication number	US-8965133-B2
Application number	US-201213646265-A
Country	US
Kind code	B2
Filing date	Oct 5, 2012
Priority date	Oct 28, 2011
Publication date	Feb 24, 2015
Grant date	Feb 24, 2015

Abstract

An image processing apparatus to obtain highly reliable local feature point and local feature amount. With the number of local feature points as a factor of an image local feature amount description size, the reproducibility of the local feature point and local feature amount is estimated, and description is made by the local amount description size, sequentially from local feature point and local feature amount with the highest reproducibility. It is possible to ensure a local feature amount description size and search accuracy.

First claim

What is claimed is:

1. An image processing apparatus for extracting, from an input image, a plurality of pairs of a local feature point in the input image and a local feature amount as a feature amount in the local feature point, comprising:
a local feature point extraction unit configured to extract a plurality of local feature points from the input image using a predetermined algorithm having robustness for image rotation;
a feature amount calculation unit configured to calculate a local feature amount, having stability against image rotation, of each local feature point;
a rotation processing unit configured to perform rotation processing at a previously set angle on said input image; and
a determination unit configured to estimate stability of local feature points and local feature amounts, obtained by applying said local feature point extraction unit and said local feature amount calculation unit to respective images before and after the rotation processing by said rotation processing unit, and to determine a predetermined number of pairs of local feature points and local feature amounts as effective local feature points and local feature amounts with respect to said input image, sequentially from a pair with a highest stability, wherein the predetermined number depends upon a resource of the image processing apparatus but does not depend upon said input image,
wherein at least one of the units is implemented by a processor.

2. The apparatus according to claim 1 , further comprising a reduced image generation unit configured to generate reduced images in a plurality of stages, by recursively performing reduction processing with a previously set reduction ratio on said input image,
wherein said rotation processing unit performs the rotation processing on each reduced image,
and wherein said determination unit determines effective local feature points and local feature amounts by each reduced image.

3. The apparatus according to claim 1 , wherein the number of effective local feature points and local feature amounts determined by said determination unit depends on a size of each reduced image.

4. A control method for an image processing apparatus for extracting, from an input image, a plurality of pairs of a local feature point in the input image and a local feature amount as a feature amount in said local feature point, comprising:
a local feature point extraction step of extracting a plurality of local feature points from the input image using a predetermined algorithm having robustness for image rotation;
a local feature amount calculation step of calculating a local feature amount, having stability against image rotation, of each local feature point;
a rotation processing step of performing rotation processing at a previously set angle on said input image; and
a determination step of estimating stability of local feature points and local feature amounts, obtained by applying said local feature point extraction step and said local feature amount calculation step to respective images before and after the rotation processing at said rotation processing step, and determining a predetermined number of pairs of local feature points and local feature amounts as effective local feature points and local feature amounts with respect to said input image, sequentially from a pair with a highest stability, wherein the predetermined number depends upon a resource of the image processing apparatus but does not depend upon said input image.

5. A non-transitory computer-readable storage medium storing a program which is read and executed by a computer, for causing said computer to execute the respective steps in the method according to claim 4 .

6. An image processing apparatus comprising:
an extraction unit to extract a plurality of feature points from an input image using a predetermined algorithm having robustness regarding the position of the feature point even if applying a predetermined change to the input image;
a feature amount calculation unit to calculate a feature amount in each feature point using a second predetermined algorithm;
an estimation unit to estimate fluctuation of the feature amount in said feature point when the predetermined change is applied to said input image; and
a determination unit to select a predetermined number of feature points having a feature amount with small fluctuation based on said estimation, and to determine the feature amounts in said selected feature points as feature amounts with respect to said input image, wherein the predetermined number depends upon a resource of the image processing apparatus but does not depend upon said input image,
wherein at least one of the units is implemented by a processor.

7. The apparatus according to claim 6 , further comprising a reduced image generation unit to generate reduced images in a plurality of stages, by recursively performing reduction processing with a previously set reduction ratio on said input image,
and wherein said determination unit selects the predetermined number of feature points and determines stabled feature amounts by each reduced image.

8. The apparatus according to claim 7 , wherein said determination unit selects the feature points by a number assigned in correspondence with size of each reduced image.

9. The apparatus according to claim 6 , further comprising a search unit to search for an image similar to the input image from images held in a holding unit using said determined feature amounts.

10. The apparatus according to claim 6 , wherein said predetermined change is rotation.

11. The apparatus according to claim 6 , wherein said fluctuation of feature amount is estimated using a Mahalanobis distance.

12. The apparatus according to claim 6 , wherein said predetermined number is set based on a data capacity of feature amounts linked to said input image.

13. The apparatus according to claim 6 , wherein said feature point is extracted using a Harris operator.

14. The apparatus according to claim 6 , wherein said feature amount is an SIFT feature amount.

15. An image processing method comprising:
an extraction step of extracting a plurality of feature points from an input image using a predetermined algorithm having robustness regarding the position of the feature point even if applying a predetermined change to the input image;
a feature amount calculation step of calculating a feature amount in each feature point using a second predetermined algorithm;
an estimation step of estimating fluctuation of feature amount in said feature point when the predetermined change is applied to said input image; and
a determination step of selecting a predetermined number of feature points having a feature amount with small fluctuation based on said estimation, and determining feature amounts in said selected feature points as feature amounts with respect to said input image, wherein the predetermined number depends upon a resource of the image processing apparatus but does not depend upon said input image.

16. A non-transitory computer-readable storage medium storing a program which is read and executed by a computer, for causing said computer to execute the respective steps in the method according to claim 15 .

Assignees

• Canon Kk

Inventors

• Shiiyama Hirotaka

Classifications

• G06F18/2113 Primary

  by ranking or filtering the set of features, e.g. using a measure of variance or of feature cross-correlation · CPC title

• G06V10/462

  Salient features, e.g. scale invariant feature transforms [SIFT] · CPC title

• G06K9/4671

  Physics · mapped topic

• G06K9/623 Primary

  Physics · mapped topic