Method and apparatus for compensating for color variation in a video system

• US 5,850,473 A
• Filed: 01/07/1997
• Issued: 12/15/1998
• Est. Priority Date: 01/07/1997
• Status: Expired due to Term

• Current Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Andersson, Russell L.
• Primary Examiner(s): Tran, Phuoc
• Application Number:
  US08/782,008
• Time in Patent Office: 707 Days
• Field of Search: 382/103, 382/162, 382/165, 382/167, 382/274, 348/578, 348/599
• US Class Current: 382/165
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/10152 Varying illumination
• CPC Class Codes: G06T 5/94 based on local image properties, e.g. for local contrast enhancement
• CPC Class Codes: H04N 23/88 for colour balance, e.g. white-balance circuits or colour temperature control

Abstract

Compensating for color variations affecting color detection by a video system. In one aspect, compensation for highlights of a prop having multiple color regions is provided. The prop is preferably designed so that every color region of the prop is affected by highlights at the same rate. The color mix ratio for a color region of the prop representing the deviation of the color region from a nominal color value is computed, and an expected color for each color region of the prop is then computed based on the color mix ratio. In another aspect, compensation for variations in a light source illuminating the environment detected by the video system are provided in combination with highlight compensation. A white color balance ratio for a frame of video data is estimated, the color ratios of background white pixels are detected and these ratios are then used to continually update the estimate of the white color balance ratios.

Claims

1. A method of tracking a moving prop having one or more color regions, each of said color regions being subject to deviation from a nominal value, comprising the steps of:

observing a color ratio of one of said color regions of said moving prop;

for each of said color regions, updating an expected color ratio based on said observed color ratio; and

employing said expected color ratio for detecting said color region of said moving prop.

2. A method of tracking a prop having one or more color regions, each of said color regions being subject to deviation from a nominal value, comprising the steps of:

observing a color ratio of one of said color regions of said prop;

for each of said color regions, updating an expected color ratio based on said observed color ratio;

employing said expected color ratio for detecting said color region of said prop; and

detecting a loss of tracking of said prop.

3. The method of claim 2 further comprising the step of employing a default observed color ratio when said loss of tracking of said prop is detected.

4. A method of tracking a prop having one or more color regions, comprising the steps of:

detecting a color ratio for a selected color region of said prop;

computing for said color region a mix ratio, said mix ratio representing a proportional deviation from a nominal color value of said color region;

computing for each of said color regions of said prop a desired color, said desired color being the sum of said nominal color and the product of said mix ratio and a difference between said nominal color and a maximum whiteness of said color region; and

employing said desired color of each of said color regions in tracking said color regions.

5. The method of step 4 and further including clipping said mix ratio to a predetermined range.

6. The method of step 5 and further including the step of detecting loss of tracking of said prop.

7. The method of step 6 and further including the step of substituting for said mix ratio a predetermined mix ratio when a loss of tracking of said prop is detected.

8. A color compensation system, comprising:

a prop having a plurality of color regions, each of said color regions being subject to whiting out at an equal rate;

a detector for detecting a color ratio of a measured color region, said measured color region being one of said color regions;

a processor for;

(i) computing for said color region a deviation from a nominal value of said color region; and

(ii) computing for each of said color regions of said prop an expected color ratio, said ratio representing said nominal color of said color region, adjusted by said deviation from said nominal value of said measured color region.

15. The color compensation system of claim 8 further comprising a white color balance compensation system for a succession of frames of video data, where the color compensation system is adapted for use in a video system, said video system also employing a color detection system and a color recognition system, the white color balance compensation system comprising:

an arithmetic unit for;

(i) receiving from said color detection system a color value for each pixel in a frame of video data;

(ii) computing a deviation ratio for each of said pixels, said deviation ratio representing a deviation from white of each of said pixels;

a first comparator for selecting pixels of interest from among said pixels, said pixels of interest being pixels for which said deviation is less than a predetermined value;

a second comparator for evaluating said pixels of interest to determine which of said pixels of interest has a deviation of less than zero and which of said pixels of interest has a deviation of more then zero;

a first counter for counting those of said pixels of interest for which said deviation is less than zero;

a second counter for counting those of said pixels of interest for which said deviation is greater than zero;

a processor for;

(i) using said counts of said pixels to construct an error function;

(ii) using said error function to update said estimate of said color balance ratio;

(iii)providing said updated estimate of said color balance ratio to said color recognition system.

9. A method of white balance compensation for a succession of frames of video data, comprising the steps of:

for each frame, estimating a white color balance ratio;

for each frame, also estimating color ratios for background white pixels;

employing said estimates of said color ratios to compute a new estimate of said white color balance ratio for each subsequent frame of video data; and

also employing said estimates of said color ratios to compensate for variations in a light source affecting subsequent frames of video data.

10. A method of claim 9 further comprising a method for white balance compensation for a succession of frames of video data, each of said frames consisting of pixels, each of said pixels having a color, comprising the additional steps of:

for a first frame of video data, estimating a white color balance ratio for said first frame of video data;

establishing a criterion by which each of said pixels will be determined to be or not to be a background white pixel;

sequentially detecting said color of each of said pixels;

evaluating, for each pixel, said color of said pixel to determine if said pixel is a background white pixel;

employing said colors of said background white pixels to update said estimated white color balance ratio;

employing said estimated white color balance ratio as a new estimated white color balance ratio for determining a white color balance ratio for a subsequent frame of video data; and

employing said estimated white color balance ratio to compensate for variations in a light source affecting said frame of video data.

11. The method of claim 10 where the step of evaluating, for each pixel, said color of said pixel to determine if said pixel is a background white pixel also includes the steps of:

computing, for each of said pixels, a deviation ratio, said deviation ratio representing a deviation from white of each of said pixels;

comparing said deviation ratio of each of said pixels against a predetermined value; and

selecting as a background white pixel each of said pixels for which said deviation ratio is less than said predetermined value.

12. The method of claim 11, where the step of employing said colors of said background white pixels to update said estimated white color balance ratio further includes the steps of:

counting all of said background white pixels for which said deviation is less than zero;

counting all of said background white pixels for which said deviation is greater than zero;

employing said counts of said background white pixels to compute a ratio to be used as an error function; and

employing said error function to drive a servo loop to update said white color balance ratio.

13. The method of claim 12, wherein said pixels to be evaluated are restricted to a predetermined range of luminance values, and wherein said step of computing a deviation ratio for each of said pixels includes precomputing a product of said estimated white color balance ratio and an average of said predetermined range of luminance values.

14. The method of claim 12, wherein said white color balance ratio is estimated over two frames of video data, said counts of said background white pixels being performed with respect to red color values of said pixels in one frame and with respect to blue color values of said pixels in an alternate frame.