Method and system for field sequential color image capture

Details Method and system for field sequential color image capture Method and system for field sequential color image capture

1999-05-06

2004-02-24

Garber, Wendy R. (Department: 2612)

Television

Camera, system and detail

With single image scanning device supplying plural color...

C348S266000, C348S273000

Reexamination Certificate

active

06697109

ABSTRACT:

FIELD OF THE INVENTION
This invention relates to the capture of color motion imagery by using sequential frame samples of different color filtration, referred to as field sequential color capture.
BACKGROUND OF THE INVENTION
Color lag is the result of capturing color components of an image at different times during exposure of a single frame. One way to solve this problem is to use multiple sensors, each capturing a color component. As will be explained, this approach has its own set of problems. There are, in the field of visual system modelling, three basic visual properties relating the luminance channel, as represented by video Y, and the opponent color channels, as approximated by the color difference signals, U and V. where B−Y=U and R−Y=V. These are:
1. The maximum temporal frequency response of the opponent color system is less than ½ that of the luminance.
2. The maximum spatial frequency response of the opponent color system is near ½ that of the luminance.
3. The maximum sensitivity of opponent color system is slightly greater than ½ that of the luminance.
The first and second properties are best described in A. B. Watson, Perceptual-components architecture for digital video, JOSA A V. 7 # 10, pp. 1943-1954, 1990; while the third property is described in K. T. Mullen. The contrast sensitivity of human colour vision to red-green and blue-yellow chromatic gratings J. Physiol. V. 359, pp. 381-400, 1985. These three properties will be used in the selection of the relative durations of exposures in order to prevent color lag and in the spatio-temporal integration following image capture.
L. Arend et al.,
Color breakup in sequentially scanned LCDs
. SID 94 Digest, pp 201-204, 1994; and D. Post et al. Predicting color breakup on field-sequential displays, SPIE Proc. V. 3058, pp 57-65, 1997, specifically investigate color break-up, or color lag. Although the research behind these papers is intended for application to field sequential color displays, some of the findings are relevant to color field sequential capture (FSC). One of these is that color lag detection has a strong luminance component. In fact, as the color offset between R, G and B layers is increased from zero to detectability, the first signal to exceed threshold is the luminance. This manifests itself as blurring in texture before any high contrast edge blur occurs, because of the masking effects caused by high contrast edges. Eventually, as the offset is increased, color artifacts become visible at high contrast edges.
Hunt,
The reproduction of colour in photography, printing, and television
, 4th edition, pp. 409-410, Fountain Press, England, describes work done with YRB camera systems. These were primarily investigated for 3-sensor systems when the sensors where tubes. The YRB system attempted to reduce the visibility of the color offset problem due to manufacturing tolerances by optically aligning the tubes. Due to complications with gamma correction, it was abandoned in favor of a RGBY systems using four tubes. These complications, which were extremely difficult to resolve in analog systems, are now easily resolved in digital signal processing.
In order to appreciate the advantages of field sequential color capture, two other common approaches must be understood. One other approach is the use of three sensors, typically using red, green, and blue filtration, which simultaneously capture the scene's dynamic content. This technique is used with both tube pickup devices and with 2D sensor arrays, such as charge coupled devices (CCD) or composite metal-on-silicon (CMOS) devices, which are referred to in the literature as 3-CCD cameras.
Another approach is the use of a single 2D sensor having color filtration applied separately to each pixel. The colors are arranged in spatially varying manners designed to provide a high spatial bandwidth for luminance or green, and to minimize color aliasing artifacts. The result is that each color layer has incomplete samples per frame, however, special interpolation algorithms are used to reconstruct full dimensioned frames for each color layer. This approach is known as color filter array (CFA) camera capture.
The 3-CCD, or 3-tube, approach's chief disadvantage is the cost of three sensors. A second disadvantage is the problem of color mis-registration between the three sensors due to their alignment in the optical path relative to the scene, which may impose tight manufacturing tolerances that increase manufacturing costs. Color mis-registration may cause luminance blur in textures with very small amounts of mis-registration, and may cause color bleeding, also referred to as color fringing, at both achromatic and chromatic edges. If the registration is well aligned, the 3-CCD approach achieves the resolution of the sensors for all three color layers of a frame. Due to cost issues and manufacturing constraints, however, these approaches are only used for high-end studio video cameras, and digital still cameras aimed for the professional and advanced hobbyist.
The CFA approach is less costly because it uses only a single sensor, however, its disadvantages include reduced spatial resolution, the necessity for an interpolation algorithm to reconstruct the three color frames for display, and necessity for an anti-aliasing filter to prevent diagonal luminance high spatial frequencies from aliasing into lower frequency color patterns and to prevent color high spatial frequencies from aliasing into luminance or color patterns. Consequently, there is a trade-off between sharpness and color artifacts. These artifacts are quite noticeable in such common image content as highlight reflections in eyes, and the expected luminance high spatial frequencies in texture, e.g., hair, or geometric patterns. In current implementations, fairly complex interpolation algorithms, which include pattern recognition, are used to try to maximize sharpness and minimize color spatial artifacts. The sharpness/aliasing tradeoff may be described via
FIG. 3
, using either filter
42
or filter
44
. Either filter may be increased in bandwidth by scaling their shape on the frequency axis. Though the image will appear to be sharper, signals having a frequency higher than that of the digital Nyquist will be aliased, i.e., the signal will fold over to frequencies lower than that of the digital Nyquist. These false lower frequencies are, in effect, added to the true signal. If all of the aliasing is removed, however, the image will appear blurred. It therefore requires a certain amount of craftsmanship in designing the filters to provide an appropriate amount of aliasing in order to maintain the bandwidths as high as possible. Most camera manufacturers opt to avoid any chromatic aliasing, because it is a new categorical distortion, and favor the sharpness reduction, which is already present to some extent. In summary, the two common approaches do not achieve the resolution of their sensors dimensions, either in luminance or in color.
The foregoing characteristics are depicted in
FIG. 1-6
. In
FIG. 1
, the Nyquist folding frequencies for the typical 2D CFA, also known as a Bayer pattern, are depicted, which are collectively referred to as the Nyquist boundary. Line
22
, with a slope of −1, is the boundary for green. The green boundary may achieve the value of 0.5 cycles/pixel (cy/pix) only at the horizontal and vertical frequencies. The red and blue components are shown by line
24
and are restricted to the square region limited by 0.25 cy/pix. Generally these color signals are manipulated so that G captures luminance, and R and B capture chrominance difference signals, by taking their difference or ratio with G. This means that luminance information with higher frequencies than those indicated in triangles
26
,
28
, which will alias into lower frequencies, showing up as both luminance and chrominance alias patterns. Similar effects occur for chrominance information outside of the smaller Nyquist square
24
boundary. Because it is most

Affiliated with

Also associated with

Rating

Method and system for field sequential color image capture 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 and system for field sequential color image capture, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for field sequential color image capture will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3323192