IMAGE FEATURE IDENTIFICATION AND MOTION COMPENSATION APPARATUS, SYSTEMS, AND METHODS

• US 20080226184A1
• Filed: 06/06/2007
• Published: 09/18/2008
• Est. Priority Date: 03/14/2007
• Status: Active Grant

• Current Assignee: Aptina Imaging Corporation (ON Semiconductor Corporation)
• Original Assignee: Micron Technology, Inc.
• Inventors: Lin, Sheng, Li, Dong, Silverstein, Amnon
• Granted Patent: US 7,599,566 B2
• Time in Patent Office: Days
• US Class Current: 382/238
• CPC Class Codes: H04N 23/73 by influencing the exposure time

Abstract

Apparatus, systems, and methods disclosed herein may estimate the magnitude of relative motion between a scene and an image capture device used to capture the scene. Some embodiments may utilize discrete cosine transform and/or Sobel gradient techniques to identify one or more blocks of pixels in an originating calibration image frame. Matching blocks of pixels may be located in a successive calibration image frame. Motion vectors originating at one calibration frame and terminating at the other calibration frame may be calculated. The magnitude of relative motion derived thereby may be used to adjust image capture parameters associated with the image capture device, including exposure settings.

Claims

1. An apparatus, comprising:

a gradient vector calculator to calculate a gradient vector for each pixel within a block of pixels excluding pixels located along a periphery of the block of pixels to provide a plurality of gradient vectors; and

gradient vector logic operatively coupled to the gradient vector calculator to perform a gradient vector analysis on the plurality of gradient vectors to derive a value of an image feature strength metric associated with the block of pixels.

2. The apparatus of claim 1, wherein the gradient vector calculator is configured to perform Sobel operations on the each pixel within the block of pixels to obtain the gradient vector.

3. The apparatus of claim 1, further comprising:

X-Sobel logic coupled to the gradient vector calculator to mathematically convolve a matrix of pixel luminance values comprising a subject pixel and eight pixels immediately adjacent to the subject pixel by an X-Sobel matrix to obtain an X-Sobel value.

4. The apparatus of claim 3, further comprising:

Y-Sobel logic coupled to the gradient vector calculator to convolve the matrix of pixel luminance values comprising the subject pixel and the eight pixels immediately adjacent to the subject pixel by a Y-Sobel matrix to obtain a Y-Sobel value.

5. The apparatus of claim 4, further comprising:

pythagorean logic coupled to the gradient vector calculator to perform a Pythagorean operation on the X-Sobel value and the Y-Sobel value to obtain a value of the gradient vector associated with the subject pixel; and

arctangent logic coupled to the gradient vector calculator to perform an arctangent operation on the X-Sobel value and the Y-Sobel value to obtain an angle associated with the gradient vector.

6. The apparatus of claim 1, further comprising:

a gradient vector buffer coupled to the gradient vector calculator to store the gradient vector for each pixel within the block of pixels excluding the pixels located along the periphery of the block of pixels.

7. The apparatus of claim 1, further comprising:

gradient vector threshold logic coupled to the gradient vector logic to discard ones of the plurality of gradient vectors smaller than a threshold magnitude.

8. The apparatus of claim 7, further comprising:

gradient vector rotational complement logic coupled to the gradient vector logic to rotate by 180 degrees remaining ones of the plurality of gradient vectors falling below an X axis.

9. The apparatus of claim 8, further comprising:

a gradient vector quantizer coupled to the gradient vector logic to quantize the remaining ones of the plurality of gradient vectors into bins comprising ranges of vector angles.

10. The apparatus of claim 9, further comprising:

a base gradient vector filter coupled to the gradient vector logic to discard remaining ones of the plurality of gradient vectors from a bin containing more of the remaining ones of the plurality of gradient vectors than any other bin and from bins immediately adjacent to the bin containing more of the remaining ones of the plurality of gradient vectors than any other bin to leave a set of final gradient vectors in at least one bin.

11. The apparatus of claim 10, further comprising:

a remaining gradient vector summation module coupled to the gradient vector logic to count a number gradient vectors in the set of final gradient vectors to obtain the value of the image feature strength metric associated with the block of pixels.

12. A system, comprising:

an image sensor array (ISA) to detect luminance incident to a set of ISA elements and to transform the luminance to create a set of luminance values associated with a pixel representation of an image;

a gradient vector calculator coupled to the ISA to calculate a gradient vector for each pixel within a block of pixels excluding pixels located along a periphery of the block of pixels to provide a plurality of gradient vectors; and

gradient vector logic operatively coupled to the gradient vector calculator to perform a gradient vector analysis on the plurality of gradient vectors to derive a value of an image feature strength metric associated with the block of pixels.

13. The system of claim 12, incorporated into a digital camera.

14. The system of claim 12, incorporated into a digital imaging software product.

15. The system of claim 12, incorporated into a vehicle collision avoidance system.

16. The system of claim 12, further including:

target matching logic operatively coupled to the gradient vector logic to correlate a first image feature block of pixels as selected from a first image of a scene to a second image feature block of pixels as searched for in a second image of the scene, wherein the second image of the scene is captured after capturing the first image of the scene.

17. The system of claim 12, further comprising:

motion priority logic operatively coupled to the gradient vector logic to estimate a magnitude of relative motion between a scene and an image capture device used to capture the scene, the relative motion occurring during a period of time between capturing a first image of the scene and capturing a second image of the scene.

18. The system of claim 17, further comprising:

motion vector logic coupled to the motion priority logic to calculate a magnitude of a motion vector associated with a first image feature block of pixels selected from the first image of the scene, the magnitude calculated as a distance between the first image feature block of pixels and a second image feature block of pixels as found in the second image of the scene, the distance divided by the period of time.

19. A method, comprising:

performing a gradient vector analysis on a block of pixels selected from an image to derive a value of an image feature strength metric associated with the block of pixels.

20. The method of claim 19, further comprising:

calculating a gradient vector for each pixel within the block of pixels excluding pixels located along a periphery of the block of pixels to provide a plurality of gradient vectors; and

discarding ones of the plurality of gradient vectors smaller than a threshold magnitude.

21. The method of claim 20, further comprising:

rotating by 180 degrees remaining ones of the plurality of gradient vectors falling below an X axis; and

quantizing the remaining ones of the plurality of gradient vectors into bins comprising ranges of vector angles.

22. The method of claim 21, further comprising:

discarding all of the remaining ones of the plurality of gradient vectors from a bin containing more of the remaining ones of the plurality of gradient vectors than any other bin and from bins immediately adjacent to the bin containing more of the remaining ones of the plurality of gradient vectors than any other bin to leave a set of final gradient vectors in at least one bin; and

counting a number gradient vectors in the set of final gradient vectors to obtain the value of the image feature strength metric associated with the block of pixels.

23. The method of claim 20, wherein the gradient vector is calculated according to a Sobel technique.

24. The method of claim 23, further comprising:

performing an X-Sobel operation on a subject pixel luminance matrix to obtain an X-Sobel value;

performing a Y-Sobel operation on the subject pixel luminance matrix to obtain a Y-Sobel value;

performing a Pythagorean operation on the X-Sobel value and the Y-Sobel value to obtain a value of a gradient vector associated with the subject pixel; and

performing an arctangent operation on the X-Sobel value and the Y-Sobel value to obtain a direction of the gradient vector associated with the subject pixel.

25. The method of claim 24, further comprising:

configuring the subject pixel luminance matrix to correspond to pixel luminance values associated with a subject pixel and eight pixels immediately adjacent to the subject pixel in a 3×

3 subject pixel matrix.

26. The method of claim 24, wherein the X-Sobel operation comprises:

convolving the subject pixel luminance matrix by an X-Sobel matrix to obtain the X-Sobel value.

27. The method of claim 24, wherein the Y-Sobel operation comprises:

convolving the subject pixel luminance matrix by a Y-Sobel matrix to obtain the Y-Sobel value.