Image analysis – Applications – Target tracking or detecting
Reexamination Certificate
2001-03-30
2004-07-06
Johns, Andrew W. (Department: 2621)
Image analysis
Applications
Target tracking or detecting
Reexamination Certificate
active
06760465
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates generally to the field of computer vision, and more specifically to a system, method and apparatus for detecting and tracking a selected object in a video sequence.
2. Discussion of the Related Art
Computer vision systems are known in the art. Such systems may track objects through a series of digital frames. However, many of the presently utilized systems track images only in the red-green-blue (“RGB”) colorspace. Such systems are poor at tracking objects through frames in which lighting conditions are changing.
Digital images include at least one picture element (“pixel”). Pixels are the small discrete elements that together constitute digital images. Each pixel of a digital image may be displayed on a computer monitor, or the like. Each pixel may be classified according to the amount of each of the primary colors of visible light—red, green and blue—(the “RGB colorspace”) that are present in the pixel. If 8 bits of information are used to represent the amount of light for each of the primary colors for each pixel, then with respect to the red component of an RGB image, the brightest red would be represented by the number 255 (in binary, 11111111) and a complete absence of red would be represented by the number 0 (in binary, 00000000). The amounts of green and blue in the pixel are also represented in a similar way.
However, the amounts of red, green and blue in an image represented in the RGB colorspace may change in different lighting conditions. For example, in a digital photograph of a red sweater, the red component of the RGB colorspace might have a level of “110” in medium lighting, “200” in bright lighting conditions, and “40” in dim lighting, even though the sweater has not been altered—only the lighting has changed. Therefore, since each of the RGB components are influenced by lighting conditions, it is problematic to keep track of a colored object in the RGB colorspace.
Another colorspace is the Hue-Saturation-Value (HSV) colorspace. The HSV colorspace, in constrast to the RGB colorspace, better represents what humans see. In the HSV colorspace, each pixel may be classified according to its Hue, the Saturation of its Hue, and the brightness (Value) in a pixel. Hue represents the wavelength of light present in the pixel. In the HSV colorspace, each of the visible colors of light is represented. Each pixel of an image has a Hue represented by cylindrical coordinates between 0° and 359°. Red is represented by coordinates around 0°. Yellow is represented by coordinates around 60°. Blue is represented by coordinates around 240°. Green is represented by coordinates around 300°.
Saturation represents the amount of Hue present in a pixel. If Saturation is represented on a scale between 0 and 1, a Saturation of 0.5 for a red Hue would be a medium red. A “very red” pixel would be represented by a Saturation of close to 1. A very red pixel would have so much red that it would, in fact, appear to be glowing red. A pixel with a red Hue that is not very red would be represented by a Saturation close to 0. Hues with Saturations close to zero appear to be mostly gray with only a slight amount of that Hue present.
Value is utilized to represent the amount brightness in the pixel. Value is typically represented on a scale from 0 to 1, with 1 representing the greatest amount of brightness, and 0 representing the least amount of brightness. Pixels with brightness near 0 are very dark—almost black. Pixels near 1 are very bright—almost white. If the Saturation is 0, then Value by itself represents the grayscale.
Object tracking systems in the art are deficient in that they are typically only able to accurately track objects under well-known conditions, such as within a range of illumination and with constraints on the fidelity of the camera.
Many current tracking systems convert a colored image into a binary image, the binary image being an image in which each pixel is represented by a “1” or “0”. Each “1” represents a pixel that might be a part of the object to be tracked. Such systems utilize processes to find the largest connected-object within the binary image, the largest connected-object being determined to be the tracked object. Such algorithms are very time-comsuming and generally inefficiently utilize system resources.
REFERENCES:
patent: 5016173 (1991-05-01), Kenet et al.
patent: 5265173 (1993-11-01), Griffin et al.
patent: 5459793 (1995-10-01), Naoi et al.
patent: 5644386 (1997-07-01), Jenkins et al.
patent: 5649021 (1997-07-01), Matey et al.
patent: 5659490 (1997-08-01), Imamura
patent: 5751450 (1998-05-01), Robinson
patent: 5912980 (1999-06-01), Hunke
patent: 5961571 (1999-10-01), Gorr et al.
patent: 6014167 (2000-01-01), Suito et al.
patent: 6075557 (2000-06-01), Holliman et al.
patent: 6118887 (2000-09-01), Cosatto et al.
patent: 6148092 (2000-11-01), Qian
patent: 6181817 (2001-01-01), Zabih et al.
patent: 6188777 (2001-02-01), Darrell et al.
patent: 6215893 (2001-04-01), Leshem et al.
patent: 6292575 (2001-09-01), Bortolussi et al.
patent: 6332033 (2001-12-01), Qian
patent: 6363160 (2002-03-01), Bradski et al.
patent: 6389155 (2002-05-01), Funayama et al.
patent: 6394557 (2002-05-01), Bradski
patent: 6404900 (2002-06-01), Qian et al.
patent: 6419638 (2002-07-01), Hay et al.
patent: 6445810 (2002-09-01), Darrell et al.
patent: 6483445 (2002-11-01), England
patent: 6556708 (2003-04-01), Christian et al.
patent: 2001/0042081 (2001-11-01), MacFarlane et al.
Frazer Matthew E
McVeigh Jeffrey S
Intel Corporation
Nakhjavan Shervin
Pillsbury WInthrop LLP
LandOfFree
Mechanism for tracking colored objects in a video sequence does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Mechanism for tracking colored objects in a video sequence, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Mechanism for tracking colored objects in a video sequence will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3190840