Edge detection using multiple color channels

• US 9,355,457 B1
• Filed: 04/15/2015
• Issued: 05/31/2016
• Est. Priority Date: 04/15/2015
• Status: Active Grant

• Current Assignee: Adobe Inc.
• Original Assignee: Adobe Systems Incorporated (Adobe Inc.)
• Inventors: Hadap, Sunil, Kim, Byungmoon
• Primary Examiner(s): Alavi, Amir
• Application Number:
  US14/687,396
• Time in Patent Office: 412 Days
• US Class Current: 1/1
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 7/90 Determination of colour characteristics

Abstract

Techniques are provided for detecting edges in a color image by analyzing variations in multiple color channels. No greyscale conversion is performed, thereby preserving color differences which might be lost in greyscale conversion. Gradients in multiple color channels are merged to compute a single edge snapping vector field from which edge locations are derived. By retaining the multidimensional characterization of the colors that comprise the processed image, a maximum color difference can be calculated using parameters which closely correspond to human perceptions of color. Maximum color difference is formulated using an LUV color model. The Euclidian distance between two pixels characterized using a multidimensional color model more closely approximates a perceived color difference between as compared to the Euclidian distance between the same pixels characterized using a greyscale color model. An edge snapping vector field is derived from these Euclidian distances, thereby providing a more accurate basis for edge detection.

Claims

1. A computer-implemented method for detecting edges in a multicolor digital image, the method comprising:

receiving a source image, wherein the source image is a multicolor digital image that comprises a plurality of pixels, each one of the plurality of pixels having a color;

characterizing each of the plurality of pixels in terms of three color parameters that define the color of the pixel;

determining an edge snapping vector field for the source image based on the three color parameters, the edge snapping vector field comprising a direction of maximum color difference and a magnitude of color difference in the direction; and

locating an edge in the source image based on the edge snapping vector field, wherein the edge passes through a point where the magnitude of color difference is greater than it is at neighboring points.

2. The computer-implemented method of claim 1, further comprising performing an image manipulation process that modifies a visual appearance of the located edge in the source image.

3. The computer-implemented method of claim 1, wherein:

the three color parameters are defined according to an LUV color model, and are represented by (L, u, v);

a color vector C(x) defines the three color parameters at a pixel located at position x=(x,y) in the source image; and

the edge snapping vector field is determined based on an eigendecomposition of J^TJ,

4. The computer-implemented method of claim 1, wherein characterizing each of the plurality of pixels in terms of the three color parameters that define the color of the pixel comprises storing a plurality of received color parameters in a memory device.

5. The computer-implemented method of claim 1, wherein:

the source image forms a selected portion a larger digital image; and

the source image is defined based on user input that identifies the selected portion.

6. The computer-implemented method of claim 1, wherein the source image is a digital photograph.

7. The computer-implemented method of claim 1, wherein the three color parameters are defined according to an LUV color model.

8. The computer-implemented method of claim 1, wherein determining the edge snapping vector field does not include converting the multicolor digital image into a greyscale image.

9. The computer-implemented method of claim 1, wherein the edge is oriented perpendicular to the direction of maximum color difference.

10. An edge detection system comprising:

a processor;

a memory;

a user interface module configured to receive a source image and display edge data associated with the source image, wherein the source image is a multicolor digital image that comprises a plurality of pixels, each one of the plurality of pixels having a color that is defined in terms of three color parameters; and

an edge detection module configured to generate the edge data, wherein the edge data establishes a location of a detected edge in the source image, wherein the edge detection module uses the processor to determine an edge snapping vector field for the source image based on the three color parameters, and wherein the location of the detected edge is based on a pixel in the edge snapping vector field where a magnitude of color difference is greater than it is at neighboring pixels;

wherein at least one of the user interface module and the edge detection module are further configured to store the edge data in the memory.

11. The edge detection system of claim 10, wherein the edge data stored in the memory includes metadata that defines an edge strength of the detected edge, the edge strength corresponding to the magnitude of color difference at the location of the detected edge.

12. The edge detection system of claim 10, wherein:

the three color parameters are defined according to an LUV color model, and are represented by (L, u, v);

a color vector C(x) defines the three color parameters at a pixel located at position x=(x,y) in the source image; and

the edge snapping vector field is determined based on an eigendecomposition of J^TJ,

13. The edge detection system of claim 10, further comprising an image manipulation module configured to manipulate an appearance of the source image by applying a visual effect to the location of the detected edge in the source image.

14. The edge detection system of claim 10, wherein the edge detection module is further configured to derive an edge snapping force field from the edge snapping vector field, the edge snapping force field pointing toward opposite sides of the detected edge.

15. The edge detection system of claim 10, wherein the source image is not converted to a greyscale image as a part of generating the edge data.

16. The edge detection system of claim 10, wherein:

the edge detection module is further configured to generate a color vector field that defines the three color parameters at each of the plurality of pixels; and

the three color parameters are defined according to an LUV color model.

17. A non-transitory computer readable medium having instructions encoded thereon that, when executed by one or more processors, causes an edge detection process to be carried out, the edge detection process comprising:

receiving a source image, wherein the source image multicolor digital image that comprises a plurality of pixels, each one of the plurality of pixels having a color;

characterizing each of the plurality of pixels in terms of three color parameters that define the color of the pixel;

determining an edge snapping vector field for the source image based on the three color parameters, the edge snapping vector field comprising a direction of maximum color difference and a magnitude of color difference in the direction; and

locating an edge in the source image based on the edge snapping vector field, wherein the edge is located at a point where the magnitude of color difference is greater than it is at neighboring points.

18. The non-transitory computer readable medium of claim 17, wherein locating the edge further comprises:

deriving an edge snapping force field from a rate of change of the maximum color difference along the direction of maximum color difference; and

determining a location where the edge snapping force field is zero.

19. The non-transitory computer readable medium of claim 17, wherein the edge detection process further comprises:

generating edge data characterizing the point where the magnitude of color difference is greater than it is at neighboring points; and

saving the edge data as metadata that further characterizes an edge strength at the point, wherein the edge strength at a particular location corresponds to the magnitude of color difference perpendicular to the edge at the particular location.

20. The non-transitory computer readable medium of claim 17, wherein determining the edge snapping vector field based on the three color parameters does not include converting the multicolor digital image into a greyscale image.