Apparatus and method for extracting edges and lines

• US 5,081,689 A
• Filed: 03/27/1989
• Issued: 01/14/1992
• Est. Priority Date: 03/27/1989
• Status: Expired due to Fees

• Current Assignee: Hughes Aircraft Company (Rtx Corporation)
• Original Assignee: Hughes Aircraft Company (Rtx Corporation)
• Inventors: Meyer, Robert H., Tong, Kevin K.
• Primary Examiner(s): Boudreau, Leo H.
• Application Number:
  US07/328,919
• Time in Patent Office: 1,023 Days
• Field of Search: 382/22, 382/27, 382/21, 382/56
• US Class Current: 382/199
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/30181 Earth observation
• CPC Class Codes: G06T 7/12 Edge-based segmentation

Abstract

An apparatus and method for processing image data to derive edges and line segments. Image intensity data is processed in one dimension along a series of scan lines to produce an intensity curve (10). Pairs of points along a scan line curve (10) representing curvature extrema (20) are checked to determine if the intensity difference between the curvature extrema (20) are characteristic of edges. Edge points are then determined between pairs of curvature extrema. Edge points found between curvature extrema (20) are then correlated with other edge points in previous scan lines to determine if they fall within a range of predicted line segments. Line segment data may then be processed at higher levels to identify objects in the data.

Claims

1. An apparatus for enhancing an image containing features, said image represented by pixel data, including feature as well as non-feature pixel data, said enhancement being produced without prior knowledge of said features, said apparatus comprising:

sensor unit for converting light energy from discrete pixels in said image into electrical signals, said pixels having defined locations in the X-Y plane;

preprocessor unit for digitizing said signals;

memory unit for storing said digitized signals;

means coupled to said memory unit for detecting and measuring the intensity of said signals at discrete pixels along a plurality of scan lines throughout said image, said intensity values creating an intensity curve along the scan lines;

means for identifying curvature extrema as those points on said intensity curves where the intensity undergoes relatively large changes with respect to previous and subsequent pixels;

means for measuring the difference in intensity between pairs of curvature extrema and identifying those pairs of curvature extrema for which said difference exceeds a predetermined threshold;

means for identifying individual edge points as a point along the scan line between said identified pairs for curvature extrema;

means for identifying first and second edge points along first and second neighboring scan lines;

means for extrapolating a line segment between said first edge point and said second edge point if the second edge point lies within n pixels of said first edge point, where n is an integer;

means for calculating upper and lower line direction limits by extrapolating a line from said first edge point to a point X pixels above said second point, where X is an integer, and by extrapolating a line from said first edge point to a point X pixels below said second point;

means for extrapolating a new line segment between said first edge point and a third edge point if that third edge point lies within aid upper and lower line direction limits; and

,

display unit for creating an enhanced image containing only said identified edge points and line segments, wherein said enhanced image contains predominantly edge points and lines representing said features and is relatively free of said non-feature pixels.

2. The apparatus of claim 1 wherein said scan lines are vertical as well as horizontal scan lines.

3. The apparatus of claim 1 further comprising a means for comparing said curvature extrema for each point with a second predetermined threshold that is proportional to the slope curvature for that point and means for ignoring those curvature extrema which exceed said second threshold, whereby noise in said pixel data is ignored.

4. The apparatus of claim 1 further comprising a means for comparing the average intensity change per pixel between neighboring curvature extrema and comparing this value with a third predetermined threshold wherein if said third threshold is not exceeded an edge point will not be identified between said curvature extrema.

5. A method for enhancing an image containing features, said image represented by pixel data, including feature as well as non-feature pixel data, said enhancement being produced without prior knowledge of said features, said method comprising the steps of:

converting light energy from an image into electrical signals corresponding to the intensity of said light energy in discrete pixels in said image, said pixels each having a defined location in the X-Y plane;

assigning the value of said electrical signal for each pixel to each defined location;

selecting one linear scan line of said pixels in the X-Y plane;

identifying curvature extrema in said scan line by determining those pixels for which said electrical signal values change most rapidly with respect to the values for neighboring pixels along said scan line;

measuring the difference in the magnitude of said electrical signal value between all pairs of curvature extrema along said scan line;

determining selected pairs of curvature extrema for which said difference exceeds a predetermined threshold;

identifying as an edge point a pixel along the scan line having an electrical signal value between the value for said identified pair, said pixel being located between said identified pair along said scan line; and

displaying the X-Y location of said edge point;

a) identifying first and second edge points along first and second neighboring scan lines;

b) extrapolating a line segment between said first edge point and said second point if the second edge point lies within n pixels of said first edge point, where n is an interger;

c) calculating upper and lower line direction limits by extrapolating a line from said first edge point to a point X pixels above said second point, where X is an integer, and by extrapolating a line from said first edge point to a point X pixels below said second point;

d) extrapolating a new line segment between said first edge point and third edge point if that third edge point lies within said upper and lower line direction limits;

e) calculating a new upper or lower line direction limit by extrapolating a line from said first point to a point X pixels above said third edge point and extrapolating a line from said first point to a point X pixel below said third edge point, and using only those direction limits which create a narrower range of line direciton limits;

f) repeating steps b-e for a fourth and all succeeding edge points in each succeeding scan line until no edge point is found within the curren tupper and lower direction limits;

g) repeating steps a-f for edge points that fall outside upper and lower limits to create new line segments; and

displaying an enhanced image containing only said identified edge points and line segments, wherein said enhanced image contains predominantly edge points and lines representing said features and is relatively free of said non-feature pixels.

6. The method of claim 5 wherein said step of converting light energy further comprises the step of providing a CCD sensor for converting said light energy.

7. The method of claim 5 further comprising the steps of digitizing said electrical signal into a digital signal for each pixel value.

8. The method of claim 7 further comprising the step of transmitting said digitized signals to a programmable processor and storing said intensity values in memory locations corresponding to pixel coordinates in said image.

9. The method of claim 5 wherein said step of identifying curvature extrema further comprises the step of determining those points where the derivative of the intensity curve exceeds a second threshold.

10. The method of claim 5 wherein said step of identifying as an edge point further comprises the step of determining the point having an intensity value equal to the average intensity of the identified pair of curvature extrema.

11. An apparatus for enhancing an image containing features, said image represented by pixel data, including feature as well as non-feature pixel data, said enhancement being produced without prior knowledge of said features, said apparatus comprising:

sensor unit for converting light energy from discrete pixels in an image into electrical signals, said pixels each having a defined location in the X-Y plane;

preprocessor means for digitizing and signal;

means for assigning the value of said electrical signal for each pixel to each defined location;

means for selecting one linear scan line of said pixels in the X-Y plane;

means for identifying curvature extrema in said scan line by determining those pixels for which said electrical signal changes most rapidly with respect to the electrical signal assigned to neighboring pixels along said scan line;

means for measuring the difference in the magnitude of said electrical signal value between all pairs of curvature extrema along said scan line;

means for determining selected pairs of curvature extrema for which said difference exceeds a predetermined threshold;

means for identifying as an edge point a pixel along the scan line having an electrical signal value between the value for said identified pair and located between said identified pair along said scan line;

means for identifying first and second edge points along first and second neighboring scan lines;

means for extrapolating a line segment between said first edge point and said second edge point if the second edge point lies within n pixels of said first edge point, where n is an integer;

means for calculating upper and lower line direction limits by extrapolating a line from said first edge point to a point X pixels above said second point, where X is an integer, and by extrapolating a line form said first edge point to a point X pixels below said second point;

means for extrapolating a new line segment between said first edge point and a third edge point if that third edge point lies within said upper and lower line direction limits; and

display unit for creating an enhanced image containing only said identified edge points and line segments, wherein said enhanced image contains predominantly edge ponits and lines representing said features and is relatively free of said non-feature pixels.

12. The apparatus of claim 11 wherein said means for converting light energy is a CCD imager.

13. The apparatus of claim 11 wherein said means for digitizing is a frame grabber preprocessor.

14. The apparatus of claim 11 further comprising means for comparing said curvature extrema for each point with a second predetermined threshold that is proportional to the slope curvature for that point and means for ignoring those curvature extrema which exceed said second threshold, whereby noise in said image data is ignored.