Image-feature detection

• US 9,158,991 B2
• Filed: 02/22/2013
• Issued: 10/13/2015
• Est. Priority Date: 02/22/2012
• Status: Active Grant

• Current Assignee: STMicroelectronics International N.V. (STMicroelectronics NV)
• Original Assignee: STMicroelectronics SRL (STMicroelectronics NV)
• Inventors: Pau, Danilo Pietro, Bruna, Arcangelo Ranieri
• Primary Examiner(s): FUJITA, KATRINA R
• Application Number:
  US13/773,847
• Publication Number:
  US 20130216097A1
• Time in Patent Office: 963 Days
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]
• CPC Class Codes: G06V 10/50 by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis

Abstract

An embodiment is a method for detecting image features, the method including extracting a stripe from a digital image, the stripe including of a plurality of blocks; processing the plurality of blocks for localizing one or more keypoints; and detecting one or more image features based on the one or more localized keypoints.

Claims

1. An apparatus, comprising:

analyzer circuitry configured to,

extract a stripe from a digital image, the stripe including a plurality of image blocks,

transform each of the image blocks in the stripe from a pixel domain into the frequency domain,

filter each of the transformed image blocks, and

analyze each filtered transformed image block to determine if each transformed image block likely includes one or more keypoints;

scale filter those filtered transformed image blocks that are determined to likely include one or more keypoints,

transform the scale filtered transformed image blocks back into the pixel domain;

analyze the transformed scale filtered transformed image blocks to localize one or more keypoints, and

detection circuitry configured to identify one or more image features based on the one or more localized keypoints.

2. The apparatus of claim 1 wherein the stripe represents a horizontal section of the digital image.

3. The apparatus of claim 1 wherein the stripe represents a vertical section of the digital image.

4. The apparatus of claim 1, wherein scale filtering those filtered transformed image blocks that are determined to likely include one or more keypoints comprises bandpass filtering the filtered transformed image blocks that are determined to likely include one or more keypoints.

5. The apparatus of claim 1 wherein the detection circuitry is configured to identify the image feature in response to a section of the image block, the section related to the image-feature.

6. The apparatus of claim 1 wherein the detection circuitry is configured to identify the image feature by:

determining a respective gradient of each of components of the image block, the components corresponding to a location of the keypoint within the portion of the image;

determining a respective orientation for each of the gradients; and

identifying the image feature in response to the gradients and orientations.

7. The apparatus of claim 1 wherein the detection circuitry is configured to identify the image feature by:

determining a respective gradient of each of components of the section of the image block;

determining a respective orientation for each of the gradients; and

identifying the image feature in response to the gradients and orientations.

8. A system, comprising:

an image-capture device configured to capture an image; and

an apparatus, including,

analyzer circuitry configured to extract a stripe from the captured digital image, the stripe including a plurality of image blocks and the analyzer circuitry further configured to filter each of the transformed image blocks, analyze each filtered transformed image block to determine if each transformed image block likely includes one or more keypoints, scale filter those filtered transformed image blocks that are determined to likely include one or more keypoints, transform the scale filtered transformed image blocks back into the pixel domain, and analyze the transformed scale filtered transformed image blocks to localize one or more keypoints; and

detection circuitry configured to identify an image feature using the one or more keypoints and determine locations of these keypoints.

9. The system of claim 8 wherein the image-capture device and the apparatus are disposed on a same integrated circuit.

10. The system of claim 8 wherein the image-capture device and the apparatus are disposed on respective integrated circuits.

11. The system of claim 8, further comprising a control circuit configured to control respective operations of the image-capture device and the apparatus.

12. A method of detecting features in a captured digital image, comprising:

identifying a stripe including a subset of the total number of pixels of the captured digital image;

locating a point or points of interest within an image block of the stripe, the stripe including a plurality of image blocks, and the locating including transforming the image blocks to the frequency domain;

filtering the blocks transformed to the frequency domain; and

describing an image feature or features of the captured digital image using the located point or points of interest within the image blocks; and

determining whether each image block potentially includes a point of interest based on a frequency distribution of the filtered block; and

wherein only image blocks determined to potentially include a point or points of interest are subjected to locating a point or points of interest within those image blocks.

13. The method of claim 12, further comprising describing the image feature or features in response to a spatial frequency within the image blocks.

14. The method of claim 12 wherein the identifying includes:

setting at least some components of the image block having a relationship to a threshold equal to a same value, the components of the image block corresponding to a location of the point of interest within the portion of the image;

quantizing the components of the image block after the setting; and

entropy coding the quantized components of the vector, the coded quantized components identifying the image feature.

15. The method of claim 12 wherein the describing includes:

determining a respective gradient of each of components of the image block, the components corresponding to a location of the point of interest within the portion of the image;

determining a respective orientation for each of the gradients; and

describing the image feature in response to the gradients and orientations.

16. A non-transitory computer-readable storage medium storing instructions that, when executed by a computing apparatus, cause the computing apparatus, or another apparatus under control of the computing apparatus:

to identify a stripe including a subset of the total number of pixels of the captured digital image;

to locate a point or points of interest within an image block of the stripe, the stripe including a plurality of image blocks, and the locating including transforming the image blocks to the frequency domain;

to filter the blocks transformed to the frequency domain;

to identify an image feature or features of the captured digital image using the located point or points of interest within the image blocks;

to determine whether each image block potentially includes a point of interest based on a frequency distribution of the filtered block; and

to locate a point or points of interest only within those image blocks determined to potentially include a point or points of interest.

17. The non-transitory computer-readable storage medium of claim 16, wherein the stripe is a horizontal stripe of the captured digital image.

18. The non-transitory computer-readable storage medium of claim 16, wherein each of the image blocks includes a fixed number of pixels of the captured digital image or the image blocks include varying numbers of pixels of the captured digital image.