Television – Camera – system and detail – Camera image stabilization
Reexamination Certificate
2002-09-06
2003-11-25
Garber, Wendy R. (Department: 2612)
Television
Camera, system and detail
Camera image stabilization
C348S208990, C348S169000, C382S295000
Reexamination Certificate
active
06654049
ABSTRACT:
TECHNICAL FIELD
The present invention relates to analysis of video images and more specifically to image stabilization of video images.
BACKGROUND
Video stabilization is known within the prior art to minimize image jitter due to camera movement through either mechanical feedback systems or through digital signal processing. The digital signal processing techniques of the prior art are complicated and often are based upon motion estimation and vector analysis. These techniques are designed so as to be minimally intrusive to movement within the image and assess movement of blocks of pixels in order to isolate camera movement jitter.
When objects move within captured video images as opposed to the camera moving details of the moving object being captured often are hard to discern. For example, a license plate of a moving car is often hard to read when the captured video images are displayed due to the great amount of motion from frame to frame of the car. In such a situation, an accounting for the motion of the car would enhance the ability to read the license plate.
The traditional image stabilization techniques as discussed above do not provide a way to account for motion such as the motion of a car in order to make a license plate more legible because the techniques are designed to allow such motion to be visible within the sequence of video images while only removing movement due to the movement of the camera.
SUMMARY OF THE INVENTION
In one embodiment of the invention, a method for image stabilization of at least two digital images is disclosed. In such an embodiment, a first digital image having a plurality of pixels and at least a second image having a plurality of pixels are provided. Each pixel has an associated address for display and is representative of a color. Either a user of the system provides a color match range or the system provides a predetermined color matched range. A pixel is selected within the first digital image. Preferably the pixel is one that represents an item within the image that is either blurry due to movement of the item or appears jittery due to camera movement. Due to the vast number of available colors a pixel selected in a first image can be matched to a pixel in a second image within a range. The range allows for compensation due to lighting changes. Once the pixel in the first image is selected it can be compared to all pixels within the second image. Each pixel within the second image that is within the color match range is saved and the closest pixel color to the pixel from the first image is selected. The address of the pixels in the second image are then readdressed such that the address of the pixel located in the second image that has the closest color to the pixel in the first image now has the same display address as that of the pixel in the first image. The repositioned second digital image is then stored to memory.
In other embodiments the image area that is to be searched within the second image may be selected by a user of the method. In another embodiment an alignment point can be selected for aligning the first digital image and the repositioned second digital image such that the pixel of the first frame is aligned with the alignment point and the pixel in the second digital image is aligned with the alignment point when displayed on a display device.
Once the second digital image is saved the first digital image and the repositioned second digital image may be displayed sequentially on a display device wherein the movement of the item will be minimized.
In a further embodiment, interpolation may be employed to determine color values for comparison in a sequence of video images. In the sequence, a beginning digital image and an ending digital image from the sequence creating a subset sequence are selected. A pixel is the selected in the beginning digital image and a pixel is selected in the ending digital image. Linear interpolation is performed between the color of the pixel in the beginning digital image and the color of the pixel in the ending digital image such that an interpolated color is determined for each image between the beginning digital image and the ending digital image. For each interpolated color associated with an image, a pixel is searched for within that image within a predetermined range of colors. Once a pixel is found, each image is repositioned so that the address of the pixel having the interpolated color is positioned to the address of the pixel selected in the beginning image.
In certain embodiments, the color match range may be equal to zero, such that an exact match needs to be found. The greater the number of representative colors, such as true color or 16.7 million colors, the greater the likelihood that a match of one pixel in a first frame represents the same physical location as that of a pixel in a second frame having the same or a very similar color. If a color is not found in an image that is within a color match range, the method can default to a standardized formula for repositioning the image based upon historic information.
The method may also be implemented in a computer system and embodied on a computer program product for use with the computer system in which the method is embodied in computer readable code. Further, the method may be embodied in a system which includes a user interface, a processor, and associated memory. The processor includes a user interface module for receiving signals from the user interface. In such a configuration, a user may send signals to the processor through the user interface indicating the video sequence to be processed, indicating a color match range, selecting pixel values, selecting alignment positions, for example. The processor also includes a location module for locating a pixel within a second frame having an associated color that is within a color range of a color associated with the pixel selection from the first video frame. The processor further includes a repositioning module for readdressing the second video frame so that the located pixel of the second frame has an address that is the same address as the pixel selection form the first video frame and also includes a storage module for storing the readdressed second video frame.
REFERENCES:
patent: 4864409 (1989-09-01), Platte et al.
patent: 5099323 (1992-03-01), Morimura et al.
patent: 5253052 (1993-10-01), Hanashiro et al.
patent: 5629988 (1997-05-01), Burt et al.
patent: 5796427 (1998-08-01), Suzuki
patent: 5892546 (1999-04-01), Hirabayashi et al.
patent: 5973733 (1999-10-01), Gove
patent: 6122004 (2000-09-01), Hwang
patent: 6144405 (2000-11-01), Toba
patent: 6211913 (2001-04-01), Hansen et al.
patent: 6493041 (2002-12-01), Hanko et al.
patent: 2002/0168091 (2002-11-01), Trajkovic
patent: 0 613 020 (1994-08-01), None
Guestrin, C., et al., “Fast Software Image Stabilization with Color Registration,” Intelligent Robots and Systems,IEEE/RSJ InternationalConference on Victoria, BC, Canada, Oct. 13, 1998, 19-24.
International Search Report, dated Nov. 13, 2002; received Nov. 21, 2002.
Fletcher Susan Heath Calvin
Grindstaff Gene Arthur
McKay, III Therman Ward
Bromberg & Sunstein LLP
Garber Wendy R.
Intergraph Hardware Technologies Company
Villecco John M.
LandOfFree
Method, device and computer program product for image... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method, device and computer program product for image..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method, device and computer program product for image... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3150604