System and method for detection of image edges using a polar algorithm process

• US 7,110,602 B2
• Filed: 08/21/2002
• Issued: 09/19/2006
• Est. Priority Date: 08/21/2002
• Status: Active Grant

• Current Assignee: Raytheon Company (Rtx Corporation)
• Original Assignee: Raytheon Company (Rtx Corporation)
• Inventors: Krause, Larry G.
• Primary Examiner(s): DESIRE, GREGORY M
• Application Number:
  US10/224,866
• Publication Number:
  US 20040037465A1
• Time in Patent Office: 1,490 Days
• Field of Search: 378/62, 378/117, 382/103, 382/176, 382/199, 382/261, 382/266, 382/203, 382/258
• US Class Current: 382/199
• CPC Class Codes: G06T 2207/10048 Infrared image
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06V 10/36 Applying a local operator, i.e. means to operate on image points situated in the vicinity of a given point; Non-linear local filtering operations, e.g. median filtering
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 20/13 Satellite images

Abstract

An edge detection system and method implement a recursive polar algorithm process to generate an edge image from an input image. The edge detection system may include a polar filter to recursively perform low-pass filtering along edge contours of a gradient pixel grid and generate a gradient image output. An edge thinner may be included to find gradient field peaks along a direction of gradient vectors of the gradient image output and generate thinned gradient pixels. An edge path segmenter may group the thinned gradient pixels into edge segments, and a segment classifier may determine whether the edge segments are edge paths or clutter. The edge detection system and method may improve automated image processing in medical diagnostics, reconnaissance, missile guidance, hazard or threat warning systems, security systems, access control systems (e.g., face recognition), navigation, geographic mapping, manufacturing quality inspection, robot vision, and search and rescue.

Claims

1. An edge detection system comprising:

a gradient operator to generate a gradient pixel grid from an input pixel grid, wherein the gradient operator generates each gradient pixel of the gradient pixel grid to be non-coincident with pixels of the input pixel grid;

a polar filter to perform low-pass filtering along edge contours of the gradient pixel grid and generate a gradient image output;

an edge thinner to find gradient field peaks along a direction of gradient vectors of the gradient image output and generate thinned gradient pixels;

an edge path segmenter to group the thinned gradient pixels into edge segments; and

a segment classifier to determine whether the edge segments are edge paths or clutter.

2. The system of claim 1, wherein the gradient operator generates each gradient pixel of the gradient pixel grid to be a two-element gradient vector representing orthogonal components of the gradient pixel.

3. The system of claim 1 further comprising a pre-filter to suppress noise of an input image outside an imager bandwidth, the input image comprising the input pixel grid.

4. The system of claim 3 wherein the pre-filter applies a point-spread function (PSF) to the input image, the pre-filter having substantially the same spectral response as an imager that generated the input image.

5. The system of claim 1 wherein the polar filter restores a gradient sampling grid to coincide with pixels of the input image.

6. The system of claim 1 wherein the polar filter performs the filtering recursively until a signal-to-noise ratio (SNR) improvement of the gradient image output is below a predetermined level.

7. The system of claim 1 wherein a number of iterations of filtering by the polar filter is pre-computed based on an equivalent shift variant linear spatial filter along an ideal noiseless edged to determine an upper bound on SNR improvement.

8. The system of claim 1 wherein the edge thinner retains a gradient pixel of the gradient image output when a magnitude of the gradient pixel is a local maximum along the direction of the gradient vectors, the edge thinner zeroing other gradient pixels.

9. The system of claim 1 wherein the segment classifier determines whether the edge segments are edge paths or clutter using gradient magnitudes of the edge segment and the number of pixels within the edge segment.

10. A method of generating an edge image comprising;

generating a gradient pixel from an input grid, wherein the gradient operator generates each gradient pixel of the gradient pixel grid to be non-coincident with pixels of the input pixel grid;

performing low-pass filtering along edge contours of the gradient pixel grid to generate a gradient image output;

finding gradient field peaks along a direction of gradient vectors of the gradient image output to generate thinned gradient pixels;

grouping the thinned gradient pixels into edge segments; and

determining whether the edge segments are edge paths or clutter to generate the edge image.

11. The method of claim 10 wherein a number of iterations of low pass filtering is pre-computed based on an equivalent shift variant linear spatial filter along an ideal noiseless edge to determine an upper bound on SNR improvement.

12. The method of claim 11 further comprising suppressing noise of an input image outside an imager bandwidth, wherein the input image comprises the input pixel grid,

wherein suppressing noise comprises applying a point-spread function (PSF) to the input image, the pre-filter having substantially the same spectral response as an imager that generated the input image.

13. The method of claim 12 further comprising:

retaining a gradient pixel of the gradient image output when a magnitude of the gradient pixel is a local maximum along the direction of the gradient vectors, the edge thinner zeroing other gradient pixels; and

determining whether the edge segments are edge paths or clutter using gradient magnitudes of the edge segment and the number of pixels within the edge segment squared.

14. A target acquisition system comprising:

a memory to store a model image, the model image having a target aim point;

an edge detector to process a real-time target image to form an edge image;

a scene matcher to match the edge image with the model image to locate the target aim point within the edge image; and

a scene image tracker coupled to a missile/seeker navigation system to maintain the target aim point within an aim point position,

wherein the edge detector comprises;

a polar filter to perform low-pass filtering along edge contours of a gradient pixel grid and generate a gradient image output;

an edge thinner to find gradient field peaks along a direction of gradient vectors of the gradient image output and generate thinned gradient pixels;

an edge path segmenter to group the thinned gradient pixels into edge segments; and

a segment classifier to determine whether the edge segments are edge paths or clutter.

15. The system of claim 14 wherein the edge detector further comprises a gradient operator to generate a gradient pixel grid from an input pixel grid.

16. The system of claim 15 where in the polar filter restores a gradient sampling grid to coincide with pixels of the input image.

17. The system of claim 16 wherein a number of iterations of filtering by the polar filter is pre-computed based on an equivalent shift variant linear spatial filter along an ideal noiseless edged to determine an upper bound on SNR improvement.

18. The system of claim 17 wherein the edge thinner retains a gradient pixel of the gradient image output when a magnitude of the gradient pixel is a local maximum along the direction of the gradient vectors, the edge thinner zeroing other gradient pixels.

19. The system of claim 18 wherein the segment classifier determines whether the edge segments are edge paths or clutter using gradient magnitudes of the edge segment and the number of pixels within the edge segment squared.

20. An edge detection system comprising:

a polar filter to perform low-pass filtering along edge contours of a gradient pixel grid and generate a gradient image output;

an edge thinner to find gradient field peaks along a direction of gradient vectors of the gradient image output and generate thinned gradient pixels;

an edge path segmenter to group the thinned gradient pixels into edge segments; and

a segment classifier to determine whether the edge segments are edge paths or clutter,

wherein a number of iterations of filtering by the polar filter is pre-computed based on an equivalent shift variant linear spatial filter along an ideal noiseless edged to determine an upper bound on SNR improvement.

21. A method of generating an edge image comprising;

generating a gradient pixel grid from an input pixel grid;

performing low-pass filtering along edge contours of the gradient pixel grid to generate a gradient image output;

finding gradient field peaks along a direction of gradient vectors of the gradient image output to generate thinned gradient pixels;

grouping the thinned gradient pixels into edge segments; and

determining whether the edge segments are edge paths or clutter to generate the edge image,

wherein a number of iterations of filtering by the polar filter is pre-computed based on an equivalent shift variant linear spatial filter along an ideal noiseless edged to determine an upper bound on SNR improvement.