Rich color transition curve tracking method

• US 8,526,717 B2
• Filed: 11/21/2011
• Issued: 09/03/2013
• Est. Priority Date: 05/15/2007
• Status: Expired due to Fees

• Current Assignee: Creative LifeStyles, Inc., Vision Interface Technologies, LLC
• Original Assignee: Creative LifeStyles, Inc., Vision Interface Technologies, LLC
• Inventors: Smith, Peter Harmon, Meyer, Frank, Cloyd, John Frederick, Tekkanat, Bora
• Primary Examiner(s): Bayat, Ali
• Application Number:
  US13/301,216
• Publication Number:
  US 20120076402A1
• Time in Patent Office: 652 Days
• Field of Search: 382/133, 382/162, 382164-167, 382/172, 382/190, 382/224, 382260-264, 382276-277, 348/223.1, 348/528, 348/645, 348/E9.037, 358/1.13, 358/1.9, 358/3.02, 358502-505, 358/515, 358/518, 358/520, 358/523, 358/525, 358/530
• US Class Current: 382/162
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/30176 Document
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 10/56 relating to colour

Abstract

A method of finding targets in a color image applies ratio space rich colored filtering to identify a band of pixels that form a transition zone from one rich color to another. This pixel band is reduced to a rich color transition curve. A set of transition curves with simple geometric and color properties for the entire image, can be compared to a database of target properties to identify a target in an image.

Claims

1. A method of identifying a target in an image comprising the steps of:

using a computer processor to perform the steps of;

defining a ratio color space determining the largest color component value for each pixel in an image;

testing each pixel for the presence of a rich target color;

finding at least two pixels of different rich color types separated by a tolerance; and

using the at least two pixels to locate a target in the image.

2. The method of claim 1 further comprising:

defining patches of contiguous pixels that have a common rich target color type;

finding two patches that are of two different target rich color types separated by less than an input tolerance;

for such two patches, defining a rich transition point location on the search-line at a mid-point between the two patches;

forming a transition curve by linking up all rich transition points in all search lines that are located within an input tolerance of each other with identical flanking rich target colors; and

finding transition curves from the same target by searching for transition curves that share the same flanking rich target color types.

3. The method of claim 1 further comprising:

developing a color filter including the step of setting a rich target color indicator to a non-target color type for any pixel that has all of the original color components darker than a lower threshold or all lighter than an upper limit threshold.

4. The method of claim 1 further comprising:

developing a color filter including the step of setting a rich target color indicator to black or white color type for any pixel that has all of the original color components darker than a lower threshold or all lighter than a upper limit threshold, respectively.

5. The method of claim 2 further comprising:

repeating the process along a new search line that is perpendicular to a first search line and combining the rich transition point locations to be added to the set from a previous pass search line.

6. The method of claim 1 further comprising:

developing a color filter including the step of displaying rich target color pixel sets that were used to define a transition point.

7. The method of claim 1 further comprising:

repeating the process through a sequence of 2-D images in time as one of a video and a 3-D spatial image along a search line that is perpendicular to each image and combining the transition points can form a 3-D transition surface formed by the transition points and 3-D transition curves.

8. The method of claim 1 further including:

using a color distance equation for each pixel located at I, j, the color distance equation defined in terms of ratio space color component values (c₁, c₂, . . . c_N);

A_m=POS(B_1,m*(f(c₁)−

T_1,m))+POS(B_2,m*(f(c₂)−

T_2,m)) . . . +POS(B_N,m*(f(c_N)−

T_N,m)),

where, for color component n and color filter type m, T_n,m is a color threshold, B_n,m is the Bias, f(c_n) is a function of c_n and the function POS (A) such that POS (A)=0 if A is less than or equal to 0, ELSE POS (A)=A.

9. The method of claim 1 further comprising:

creating a corresponding look-up-table for each primary color and secondary color used in a target; and

capturing an image and subtracting from each pixel in the image the bias of each camera color component applying a ratio space look-up-table to each pixel in the image for each primary and each secondary color used in the target to implement a color filter in the color ratio space to determine the presence of any of the N factorial rich color types that are used to identify the target.

10. The method of claim 1 further comprising:

using a processor to find the location of a target within an image by searching for rich color patches adjacent to each other within a tolerance.

11. The method of claim 1 further comprising:

using more sensor components beyond red, green and blue color sensors.

12. The method of claim 1 further comprising:

enhancing poor quality data streams to emphasize rich color characteristics.

13. The method of claim 1 further comprising:

setting an indicator for each pixel that identifies the type of rich target color and the presence or the lack of a rich target color.

14. The method of claim 1 further comprising:

identifying a pixel as having a non-rich color if each of its input color components is less than a black tolerance.

15. The method of claim 1 further comprising:

identifying a pixel as having a non-rich color if each of its input color components is above a white tolerance.

16. The method of claim 1 further comprising:

filtering image data by limiting image data to patches of target color that flank each side of a rich color transition point.

17. The method of claim 1 further comprising:

using transition curves to identify targets in an image.

18. The method of claim 1 further comprising:

using transition curves and database information to determine physical properties of targets in image data.

19. The method of claim 1 further comprising:

transition curves and database information to identify targets in multidimensional image data.

20. The method of claim 1 further comprising:

associating multiple adjacent curves with a given target.

21. The method of claim 1 further comprising:

associating a path and geometric properties, a speed, and acceleration of the transition curves in the sequence of video frames to trigger use of a predefined computer program and a specific data base entry to be used with the predefined computer program.