Television – Image signal processing circuitry specific to television – Chrominance-luminance signal separation
Reexamination Certificate
2001-06-28
2004-08-10
Lee, Micahel H. (Department: 2614)
Television
Image signal processing circuitry specific to television
Chrominance-luminance signal separation
C348S666000, C348S669000
Reexamination Certificate
active
06774954
ABSTRACT:
FIELD OF THE INVENTION
The invention relates to separation of luminance (Y) and chrominance (C) in a TV composite signal, particularly to Y/C separation for a TV composite signal using a comb filter bank.
BACKGROUND
A typical color TV decoder receives a composite TV signal as input. Using a comb filter, the color decoder separates Luminance (Y) and Chrominance (C) from the input signal. Next, the comb filter applies the band-pass filtering for the C signal and the band-stop filtering for the Y signal. Thirdly, the C signal is de-modulated back to the base-band region. Fourthly, a low-pass filter band-limits both the de-modulated C signal and the Y signal. Lastly, the band-limited Y and C signals are converted to Red, Green, and Blue outputs.
The composite signal allocates Y and C in the three-dimensional spectral positions in a three-dimensional spectral space. When represented in a one-dimensional or a two-dimensional subspace of the three-dimensional spectral space, the spectrum of Y and C overlap with each other. As such, one-dimensional and two-dimensional comb filters cannot separate Y and C completely. Rather, only three-dimensional comb filters can separate Y and C from a digitized composite image sequence completely. Moreover, as a genuine nonlinear characteristic of an image, edges inside image make difficult any Y/C comb filtering for completely separating Y from C or C from Y in a composite TV signal. Even worse, in an image sequence or video, motions inside image exist in a speed to any directions. Therefore, without perfect motion estimation for the motion of edges in addition to the motion of an ordinary object, the complete Y/C separation cannot be achieved by a one-dimensional or two-dimensional linear comb filter.
However, few real three-dimensional Y/C comb filters are available for a digital color decoder. Even a conventional three-dimensional comb filter operates mostly for the condition of motionless image sequences. It does not fully take advantage of the strong points of three-dimensional comb filters.
For example, most of video signal includes lots of motion contents, but a conventional three-dimensional comb filter function merely two-dimensional comb filter in the spatial domain. Specifically, conventional three-dimensional comb filters simply apply to the conditions of motionless parts in video by a go-no-go decision, wherein the temporal filter function of these three-dimensional filters is not selected in most case. Therefore, a conventional three-dimensional comb filter typically operates merely as a two-dimensional comb filter. For that reason, the conventional three-dimensional comb filter is ineffective when filtering complex motion video content. As another example, memory buffers are necessary to store data for performing the temporal filtering of a conventional three-dimensional filter. However, by operating merely as a two-dimensional comb filter most of the time, the conventional three-dimensional filter rarely use these precious memory buffers.
BRIEF SUMMARY OF THE INVENTION
A method for separating luminance (Y) and chrominance (C) of a composite television digital signal is provided. The method includes analyzing said composite signal to search for one of a plurality of pre-defined motions; in response to a pre-defined motion being detected, filtering the spectral energy of said detected pre-defined motion by a one-dimensional temporal comb filter selected from a filer bank; in response to no pre-defined motion being detected, analyzing said signal by searching for an edge in a three-dimensional sample space of said signal, wherein said sample space is spanned by a horizontal axis, a vertical axis and a temporal axis, and wherein said edge represents a motion encoded in said signal; in response to said edge being detected, selecting a filter of said filter bank in accordance with the orientation of said edge to filter the spectral energy of said motion; and in response to no edge being detected, selecting a three-dimensional spatial-temporal comb filter of said filter bank.
A method for separating luminance (Y) and chrominance (C) of a composite TV digital signal is provided. The method includes searching for motion encoded in said signal, wherein a motion is represented as an edge in a three-dimensional sample space of said signal, said sample space spanned by a horizontal axis, a vertical axis and a temporal axis; in response to detecting a first motion that matches one of a plurality of pre-defined motions, select a one-dimensional temporal filter of a filter bank to filter spectral energy corresponding to a first edge that represents said first motion in said sample space; and in response to detecting a second motion that is different from any of said n pre-defined motions, selecting a filter of said filter bank to filter spectral energy corresponding to a second edge, said filter selected according to the orientation of said second edge in said sample space.
A method for separating luminance (Y) and chrominance (C) from a composite TV digital signal is provided. The method includes searching for a plurality of pre-defined motion cases, said plurality of pre-defined motion cases comprises the motionless case; in response to finding at least one of said pre-defined motion cases, performing Y/C separation of said composite TV digital signal by selecting a one-dimensional (1D) temporal comb filter from a filter bank, said filter bank comprises a plurality of 1D temporal comb filters in one-to-one correspondence with said plurality of pre-defined motion cases; in response to finding none of said pre-defined uniform motion cases, performing Y/C separation of said composite TV digital signal by selecting according to a selection method a filter from the portion of said filter bank that comprises a 1D horizontal (H) comb filter, a 1D vertical (V) comb filter, a two-dimensional (2D) horizontal-vertical (HV) comb filter, a 2D horizontal-temporal (HT) comb filter, a 2D vertical-temporal (VT) comb filter, a 3D horizontal-vertical-temporal (HVT) comb filter.
A filter bank for a TV composite signal is provided. The filter bank includes an one-dimensional (1D) temporal comb filter adapted to perform Y/C separation for a motionless image; a plurality of 1D temporal comb filters adapted to perform Y/C separation by limiting spectral energy associated with an image motion detected having a pre-defined uniform velocity; and a plurality of two-dimensional (2D) comb filters. The 2D comb filters include a 2D spatial (HV) comb filter adapted to limit spectral energy along horizontal and vertical frequency axes of a 3D spectral space of said signal; a 2D spatial-temporal (HT) comb filter adapted to limit spectral energy along horizontal and temporal frequency axes of said 3D spectral space, and a 2D spatial-temporal (VT) comb filter adapted to limit spectral energy along vertical and temporal frequency axes of said 3D spectral space. The filter bank includes a three-dimensional (3D) spatial-temporal (HVT) comb filter adapted to limit spectral energy along horizontal, vertical and temporal frequency axes of said 3D spectral space.
A color TV decoder for a TV composite signal is provided. The color decoder includes a motion detection unit adapted to detect any of a plurality of pre-defined velocities from a set of gray level differences; and a filter bank coupled to said motion detection unit. The filter bank includes a one dimensional (1D) temporal comb filter adapted to filter a motionless image detected by said motion detection unit; a first plurality of 1D temporal comb filters adapted to filter an image motion detected by said motion detection unit as having a velocity that is equal to one of the said plurality of pre-defined velocities; a second plurality of 1D comb filters adapted to limit spectral energy along one of the three frequency axes of spectral space; a third plurality of two-dimensional (2D) comb filters adapted to filter an image by limiting spectral energy along two of said three frequency axes; and a three-dimensional (3D) comb filter a
Lee Micahel H.
Marger & Johnson & McCollom, P.C.
NDSP Corporation
LandOfFree
Apparatus and method for adaptive three dimensional... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for adaptive three dimensional..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for adaptive three dimensional... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3276030