Television – Image signal processing circuitry specific to television – Chrominance-luminance signal separation
Reexamination Certificate
2000-03-13
2001-08-21
Hsia, Sherrie (Department: 2614)
Television
Image signal processing circuitry specific to television
Chrominance-luminance signal separation
C348S669000
Reexamination Certificate
active
06278495
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates generally to an apparatus for decoding of composite video to provide high quality serial digital output and, more particularly, to a combination of unique circuits including a three-dimensional comb filter for providing better separation of composite video into luminance and chrominance components.
Composite video signals have been the standard for most video recording, production and transmission. The composite signal may conform to the National Television System Committee (NTSC) standard in the U.S., most of the Americas, Japan and many other countries, or the Phase Alternating Line (PAL) standard in most of Europe, Africa, and Asia. The composite signal consists of the luminance (Y) signal and the chrominance (C) signal which is encoded on a subcarrier and added to the luminance signal. There is a problem with separating the composite signal back into the Y and C signals because of the process of encoding chroma on the luminance signal. The luminance signal has a frequency response of near direct current (DC) to greater than 4.2 MHz in NTSC and 5.5 MHz in PAL. The chrominance is modulated on a subcarrier of 3.579545 MHz in NTSC and 4.433619 MHz in PAL using a quadrature modulation technique. The two signals are then simply added together. This places the chrominance signal in the high frequency range of the luminance pass band. Simple band split filters leave some of the chrominance signal in the luminance and some of the luminance signal in the chrominance.
Video decoders have been used for several years that use various forms of comb filters to separate the luminance and chrominance signals. Comb filters have been used to achieve full bandwidth or near full bandwidth luminance separation for video recorders, time base correctors (TBCs), video synchronizers, video monitors, Moving Picture Experts Group (MPEG) and Joint Photograhics Experts Group (JPEG) compressors and other video devices.
A comb filter is a well known technique that adds or subtracts two or more lines of video to separate the Y and C signals. Comb filters work because the subcarrier phase of every other line is inverted in NTSC signals and every second line is inverted in PAL. The V (R-Y) vector is inverted every other line in PAL resulting in an apparent 90 degree phase shift of the subcarrier per line. In the NTSC system, a simple two line comb adds two lines of video to cancel the chrominance and leave only luminance, and subtracts the two lines to cancel the luminance to extract chrominance. In the PAL system, a simple one line delay comb will separate the V vector from the composite signal requiring another step or technique to separate the U (B-Y) vector. A so-called PAL modifier is commonly used for this purpose. A two line delay is required to make a comb filter work the same in PAL as it does for NTSC. The problem with simple 2 line comb filters is they cause a vertical smear of the chrominance signal and high frequency luminance signals at vertical transitions in the video signal (e.g., U.S. Pat. No. 3,542,945 to Parker). Three line comb filters average the line above and below the current line before adding or subtracting from the current line. This helps center the smear and reduce its effects. Adaptive comb filters were invented to further minimize the smearing problem (e.g., U.S. Pat. No. 4,240,105 to Faroudja).
An adaptive comb filter switches or fades to a band split technique of separating high frequency chrominance from lower frequency luminance signals when a difference is detected between the current line and the lines that are being used to comb filter. When a difference exists between lines it means that the comb will not extract luminance and chrominance properly and will cause chroma smears, hue shifts, and/or luminance smears (loss of high frequency resolution) at these points. This effect is commonly known as a comb failure.
There are many problems in detecting a change in chrominance between lines of video. The chroma signal is inverted from line to line but the luminance edges of vertical lines are not inverted. Therefore, comparing the high frequency video between two lines at the corners of objects results in a difference signal at the horizontal edge of an object even when the color brightness, hue and saturation are the same. The technique disclosed by Faroudja filters off the high frequency chrominance signal, which also filters off the high frequency luminance, and uses the differences between lines of the remaining low frequency luminance signal to detect a comb fail. The comb fail signal is used to switch from a comb filter to a band split filter for Y/C separation. Another older technique compares the line above and below the current video line to determine comb fail of a 3 line comb filter. This system works well on multiple line vertical edges because the line to line chroma inversion is back in phase every second line. The problem with this technique is it will fail at a vertical transition for two lines causing a rounding of the comers of objects when it switches to band split separation mode at the comb failure point.
Field and frame comb filters cause motion artifacts when they fail. All adaptive techniques have the same problem when they switch to band split mode because the high frequency component contains both luminance and chrominance. This causes a loss of horizontal resolution and luma/chroma crosstalk.
Advances in comb filter design are generally made in the comb fail detection circuits. In general these improvements are an attempt to reduce the luma and chroma smear by combing when combing is appropriate and switching to band split separation when combing would produce a comb smear artifact. There is always a trade off in artifacts due to the fact that objects on the video screen are seldom perfect transitions from one easily distinguishable object to another as a cartoon would be.
In the present state of the art, the most advanced comb filter designs are described in the patents to Raby (U.S. Pat. Nos. 5,424,784 and 5,526,060). The comb filter design disclosed uses demodulated chroma rather than the low frequency luminance data or high frequency composite chroma data to determine comb fail characteristics of the video. The U and V signals can be used as is, converted to RGB (red, green and blue) or HSI (hue, saturation and intensity) to determine comb fail. The HSI signals of three lines are compared and the differences are used as inputs to a lookup table to determine the comb fail threshold. The lookup table is used to determine the ratio of the hue, saturation and intensity differences between lines to determine the comb fail threshold.
SUMMARY OF THE INVENTION
The main differences between this invention and prior art comb filters, including those disclosed by the Raby patents, are the concept and type of circuit used to determine the comb fail characteristics, the number of simultaneous taps used, the way this invention is able to continuously vary the weighting factor between filter taps instead of making a full transition away from one tap to another and the use of a noise level measurement to adapt the comb fail characteristics with noise level. The Raby invention separates the high frequency signals from three lines simultaneously, demodulates each of them separately, transforms the demodulated signals to HSI and compares the HSI signals for differences between lines to form a comb fail signal. The present invention does not use any of these operations.
This invention uses a Fast Fourier Transform (FFT) circuit, or a simple band split filter circuit, to determine important characteristics of each line of video without demodulating the signal. These circuits produce a signature signal by which each of the lines can be correlated. The signature signals on various surrounding video lines, that are of opposite subcarrier phase with the current line, are compared to the current line to determine the similarity. The result is used to determine the weighting coefficients of the surrounding lines
Curtis John W.
Lowe Virgil L.
Fortel DTV, Inc
Hsia Sherrie
Womble Carlyle Sandridge & Rice PLLC
LandOfFree
Digital comb filter for decoding composite video signals does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital comb filter for decoding composite video signals, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital comb filter for decoding composite video signals will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2499123