Patent Grant
7944508
The present invention relates in general to digital image and video signal processing and in particular to a method and system for automatically detecting and suppressing cross-luma and cross-color artifacts in a video signal.
World wide analog video standards, such as NTSC and PAL, use interlaced video formats to maximize the vertical refresh rates while minimizing the required transmission bandwidth. In an interlaced video format, a video frame includes a plurality of pixels that are arranged in a plurality of horizontal scan lines. Each frame is split into two video fields. The first of the two fields includes the pixels located in the odd numbered horizontal scan lines, while the second field includes the pixels located in the even numbered scan lines. The interlaced video fields are transmitted sequentially in temporal order to a display system, thereby minimizing the transmission bandwidth.
Typically, a component video signal includes a luma component and chroma components. The luma component represents brightness or luminance information, while the chroma components represent color information, e.g., as color differences, Cb and Cr. The luma component signal and the chroma component signals can be combined to form a composite video signal to minimize further the bandwidth requirements for transmission and to simplify transmission. Both the NTSC and PAL analog video standards support and transmit composite video signals.
In the typical composite video signal, the chroma components are quadrature-modulated. That is, one component of the chroma signal, e.g., Cb, has a different amplitude and is 90 degrees out-of-phase with the other component of the chroma signal, e.g., Cr. The rationale for modulating the chroma subcomponent signals and combining the modulated chroma signal with the baseband luma signal to form a composite video signal is based on the frequency interleaving principle, which provides that both the modulated chroma and baseband luma signal spectra contain individual spectrum lines for stationary video and that those spectrum lines of modulated chroma and baseband luma signals will interleave with each other without overlapping when an appropriate subcarrier frequency is chosen.
During transmission, the quadrature-modulated chroma signal and the baseband luma signal share a portion of the total video signal bandwidth. For example, in NTSC (M) systems, the chroma signal is modulated on a subcarrier frequency of 3.579545 MHz. The chroma signal and luma signal are intermingled within the modulated chroma signal band which extends from roughly 2.3 MHz to 4.2 MHz. In PAL (B/D/G/H/K/N) systems, the chroma signal is modulated on a subcarrier frequency of 4.43361875 MHz. The chroma signal and luma signal are intermingled within the modulated chroma signal band which extends from roughly 3.1 MHz to 5.0 MHz.
When the composite video signal is received by the display system, the signal is decoded by a decoder that separates the modulated chroma signal and the luma signal from the composite signal. One technique for separating the modulated chroma and the luma signals uses a combination of a notch filter passing the luma signal and a bandpass filter passing the modulated chroma signal. Because the filtering of the modulated chroma and the luma signals is performed only in the horizontal domain, this decoding technique is usually called notch filter luma/chroma (Y/C) separation. The modulated chroma signal is subsequently quadrature-demodulated into the two chroma components such as, for example, I and Q signals, or U and V signals. The luma component and the two chroma components can be used in a matrix computation to generate red, green, and blue (RGB) signals when the display system is a television display system.
Another technique for separating the modulated chroma signal and the luma signal uses adaptive line comb filters, i.e., comb filters in the vertical domain using line buffers. This technique is based on the premise that, in a quadrature-modulated composite video signal, the luma signal energy in the vicinity of the modulated chroma band is most probably located near harmonics of the horizontal scanning frequency, while the modulated chroma signal energy is located half horizontal scanning frequency between the luma signal energy peaks for NTSC (M) systems, and one quarter and three quarters horizontal scanning frequency between the luma signal energy peaks for PAL (B/D/G/H/K/N) systems, depending on the relationship between the subcarrier frequency and horizontal scanning frequency. According to this technique, the chroma and luma signals are adaptively averaged across successive scan lines using comb filters based on the presence of vertical transitions in the composite video signal to prevent blurring of the chroma or luma signals in the vertical domain. Because the filtering of chroma and luma signals is done in both the horizontal and vertical domains, this technique is usually called 2-D comb filter Y/C separation, and works particularly well when the luma signal energy is concentrated around harmonics of the horizontal scanning frequency or when the luma signal features are vertical or substantially vertical.
Yet another technique for separating the luma signal from the modulated chroma signal uses motion adaptive frame comb filters, i.e., comb filters in the temporal domain using frame buffers. According to this technique, when motion is detected in the temporal domain, the chroma and the luma signals are adaptively averaged across successive video frames to prevent blurring of chroma or luma signals in the temporal domain. In addition, when vertical transitions within a frame are detected, the chroma and the luma signals are adaptively averaged across scan lines using comb filters to prevent blurring of chroma or luma signals in the vertical domain. Because the filtering of chroma and luma signals is done in the horizontal, vertical, and temporal domains, this technique is usually called 3-D comb filter Y/C separation.
As stated above, the rationale for modulating the chroma components and combining the modulated chroma signal with the luma signal to form a composite video signal is based on the frequency interleaving principle. Nevertheless, the frequency interleaving principle is not applicable for moving video and/or video containing fine diagonal lines. In these cases, the modulated chroma signal spectrum and the luma signal spectrum can and often do overlap with one other. The result is that some degree of mutual interference, i.e., cross-talk, between the luma signal spectrum and chroma signal spectrum can occur.
The term “cross-color” refers to corruption of the modulated chroma signal spectrum caused by cross-talk from the high-frequency luma signal spectrum. When a composite video signal having cross-color is decoded with either a notch filter or a line comb filter, the cross-color can cause visual artifacts that can appear as a coarse rainbow pattern or random colors in image areas having dense diagonal fine lines, such as tiled rooftops, laminated fences, herringbone patterned clothing, and leafy scenery. The term “cross-luma” refers to corruption of the high-frequency luma signal spectrum caused by cross-talk from the modulated chroma signal spectrum. When a composite video signal having cross-luma is decoded with either a notch filter or a line comb filter, the cross-luma can cause visual artifacts that can appear as fine alternating dark and bright dots in image areas having abrupt chroma transitions, such as at the boundaries of contrasting colors, like blue and yellow.
While the visual artifacts caused by cross-color and cross-luma might be tolerable when displayed on a legacy television having a small screen with low brightness and low resolution, such artifacts are highly objectionable when displayed on a modern television having a large screen with high brightness and high resolution. Moreover, the cross-color and cross-luma problems are exacerbated when an advanced image scaling function, typically included in a modern television, enlarges video from a composite video source onto the large-size display with high resolution. In such modern systems, reducing or eliminating cross-color and cross-luma artifacts is highly desirable.
While decoding systems that use notch filter Y/C separation and 2-D comb filter Y/C separation techniques fail to reduce or eliminate cross-color and cross-luma artifacts, decoding systems that use 3-D comb filter Y/C separation techniques can reduce such artifacts in certain situations. For example, because filtering is performed in the horizontal, vertical, and temporal domains, the cross-color and/or cross-luma artifacts can be reduced in stationary portions of an image, such as over fine diagonal lines and along sharp vertical chroma transitions. Nevertheless, implementing a decoder that uses adaptive 3-D frame comb filters can be significantly more expensive then one that uses notch filters or line comb filters because the decoder requires motion detectors and frame buffers. Moreover, because the technique is based on motion detection in the temporal domain, actual motion in the image can be mistakenly interpreted as cross-color and/or cross-luma, and vice versa. In this case, improper filtering can itself produce serious cross-color and cross-luma artifacts.
Other systems for suppressing cross-color and/or cross-luma artifacts implement adaptive cross-color and/or cross-luma suppression techniques that operate only upon modulated chroma signals prior to demodulation. While these systems can be effective when the input signal is a composite video signal, they cannot be used to suppress cross-color and/or cross-luma artifacts in de-modulated baseband component signals. This shortcoming is significant because many modern digital devices are configured to process baseband component signals in the form of one luma (Y) signal and two color differences (Cb and Cr) signals. Such signals are received from, for example, a digital TV (DTV) tuner or a DVD player connected through a serial or parallel digital interface conveying YCbCr component signals. These baseband component signals can exhibit cross-color and/or cross-luma artifacts when the source of the content is taken from a composite video master. In this case, cross-color and/or cross-luma artifacts in the composite video master are irreversibly imprinted and any signals derived from the master necessarily inherit the cross-color and/or cross-luma errors and the associated visual artifacts.
In addition, even when the content source is not taken from a composite video master, the baseband component signals of a video signal can exhibit cross-color and/or cross-luma artifacts when the original component video signal is converted to a composite format during any stage of video processing, such as during production, distribution, transmission, and so forth. In both cases, cross-color and cross-luma detection and suppression must be performed directly on the baseband component signals in order to reduce the objectionable artifacts.
Accordingly, it is desirable to provide a method and system for detecting and suppressing the cross-color and cross-luma present in a baseband component video signal derived from a quadrature-modulated composite video signal. The system should be cost effective and should not require extensive computational and storage resources.
A method and system for automatically detecting and suppressing the cross-color and cross-luma present in a baseband component video signal is described. In one aspect, the method includes receiving component pixel data of a current pixel located in a current position in a current scan line in a current frame, the component pixel data of a first previous pixel located in the current position in the current scan line in a first frame, and the component pixel data of a second previous pixel located in the current position in the current scan line in a second frame, wherein the second frame temporally precedes the first frame, which temporally precedes the current frame. The method further includes calculating a first difference based on the component pixel data of the first previous pixel and the component pixel data of the current pixel, a second difference based on the component pixel data of the second previous pixel and the component pixel data of the first previous pixel, and a third difference based on the component pixel data of the second previous pixel and the component pixel data of the current pixel, and determining for the current pixel whether at least one of cross-luma and cross-color is present based on an absolute value of at least one of the first difference, the second difference and the third difference. A per pixel count associated with the component pixel data of the current pixel is determined based on the determined presence of at least one of cross-luma and cross-color, and the component pixel data of the current pixel is modified based on the per pixel count associated with the component pixel data of the current pixel. The method includes outputting the modified component pixel data of the current pixel as a corrected output color video signal, where the corrected output color video signal is substantially without visual artifacts caused by the at least one of cross-luma and cross-color.
In another aspect, a system for automatically detecting and suppressing cross-color and cross-luma present in a baseband component video signal includes means for receiving component pixel data of a current pixel located in a current position in a current scan line in a current frame, the component pixel data of a first previous pixel located in the current position in the current scan line in a first frame, and the component pixel data of a second previous pixel located in the current position in the current scan line in a second frame, where the second frame temporally precedes the first frame, which temporally precedes the current frame. The system further includes means for calculating a first difference based on the component pixel data of the first previous pixel and the component pixel data of the current pixel, a second difference based on the component pixel data of the second previous pixel and the component pixel data of the first previous pixel, and a third difference based on the component pixel data of the second previous pixel and the component pixel data of the current pixel, and means for determining for the current pixel whether at least one of cross-luma and cross-color is present based on an absolute value of at least one of the first difference, the second difference and the third difference. The system further includes a means for determining a per pixel count associated with the component pixel data of the current pixel based on the determined presence of at least one of cross-luma and cross-color, and a means for modifying the component pixel data of the current pixel based on the per pixel count associated with the component pixel data of the current pixel. The system further includes means for outputting the modified component pixel data of the current pixel as a corrected output color video signal, where the corrected output color video signal is substantially without visual artifacts caused by the at least one of cross-luma and cross-color.
In another aspect, a system for automatically detecting and suppressing the cross-color and cross-luma present in a baseband component video signal includes a correction unit configured for receiving component pixel data of a current pixel located in a current position in a current scan line in a current frame, the component pixel data of a first previous pixel located in the current position in the current scan line in a first frame, and the component pixel data of a second previous pixel located in the current position in the current scan line in a second frame, wherein the second frame temporally precedes the first frame, which temporally precedes the current frame. The system also includes a detection unit configured for calculating a first difference based on the component pixel data of the first previous pixel and the component pixel data of the current pixel, a second difference based on the component pixel data of the second previous pixel and the component pixel data of the first previous pixel, and a third difference based on the component pixel data of the second previous pixel and the component pixel data of the current pixel, for determining for the current pixel whether at least one of cross-luma and cross-color is present based on an absolute value of at least one of the first difference, the second difference and the third difference, and for determining a per pixel count associated with the component pixel data of the current pixel based on the determined presence of at least one of cross-luma and cross-color. The system also includes a suppression unit configured for modifying the component pixel data of the current pixel based on the per pixel count associated with the component pixel data of the current pixel, and for outputting the modified component pixel data of the current pixel as a corrected output color video signal, wherein the corrected output color video signal is substantially without visual artifacts caused by at least one of cross-luma and cross-color.
In yet another aspect, a progressive scan display system includes a signal receiving unit, a tuner box for transforming the signal into an analog signal, a video decoder for transforming the analog signal into an interlaced component video signal comprising component pixel data, and a motion adaptive de-interlacing system for converting the interlaced component video signal into a progressive component video signal. The de-interlacing system includes a cross-color and cross-luma correction unit configured for receiving component pixel data of a current pixel located in a current position in a current scan line in a current frame, the component pixel data of a first previous pixel located in the current position in the current scan line in a first frame, and the component pixel data of a second previous pixel located in the current position in the current scan line in a second frame, where the second frame temporally precedes the first frame, which temporally precedes the current frame, a detection unit configured for calculating a first difference based on the component pixel data of the first previous pixel and the component pixel data of the current pixel, a second difference based on the component pixel data of the second previous pixel and the component pixel data of the first previous pixel, and a third difference based on the component pixel data of the second previous pixel and the component pixel data of the current pixel, for determining for the current pixel whether at least one of cross-luma and cross-color is present based on an absolute value of at least one of the first difference, the second difference and the third difference, and for determining a per pixel count associated with the component pixel data of the current pixel based on the determined presence of at least one of cross-luma and cross-color. The de-interlacing system also includes a suppression unit configured for modifying the component pixel data of the current pixel based on the per pixel count associated with the component pixel data of the current pixel, and for outputting the modified component pixel data of the current pixel as a corrected output color video signal, where the corrected output color video signal is substantially without visual artifacts caused by at least one of cross-luma and cross-color. The progressive scan display system includes a display for displaying the progressive video signal.
The accompanying drawings provide visual representations which will be used to more fully describe the representative embodiments disclosed here and can be used by those skilled in the art to better understand the representative embodiments and their inherent advantages. In these drawings, like reference numerals identify corresponding elements, and:
Methods and systems for automatically detecting and suppressing cross-color and cross-luma present in a baseband component video signal are described. The following description is presented to enable one of ordinary skill in the art to make and use the invention and is provided in the context of a patent application and its requirements. Various modifications to the preferred embodiments and the generic principles and features described herein will be readily apparent to those skilled in the art. Thus, the present invention is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features described herein.
According to an exemplary embodiment, a cross-color/cross-luma (CC/CL) correction unit is configured to receive a component video signal including chroma pixel data and the luma pixel data. The CC/CL correction unit analyzes the chroma pixel data and the luma pixel data to detect cross-color and cross-luma, respectively. In one embodiment, the detection is based on differences between the pixel data of a current pixel located in a specific position in a current video frame and the pixel data of at least two other pixels located in the same position in at least two other preceding frames. When detected, the cross-color and/or cross-luma are suppressed by the CC/CL correction unit such that the resulting component video signal can be correctly processed downstream and displayed substantially without swirling rainbows, crawling dots, and other artifacts due to cross-color and cross-luma. The CC/CL correction unit can be implemented for a television, a video receiver, a video recorder, a set-top box (STB), or for any other video processing device for the production, distribution, transmission, reception, scan format conversion, and display of a baseband component video signal.
The CC/CL correction unit 130 includes a detection unit 400 that is configured to detect cross-luma and/or cross-color in a component video signal and a suppression unit 600 configured to suppress the detected cross-luma and/or cross-color. The function of each unit 400, 600 will now be described in conjunction with
Referring to
According to an exemplary embodiment, the system 100 includes means for receiving the input component pixel data of the current pixel, first previous pixel and the second previous pixel. For example, the CC/CL correction unit 130 can be configured for receiving the aforementioned pixel data. In one embodiment, the component pixel data of the first previous pixel and the second previous pixel are stored in the storage buffer 120, and the data buffer controller unit 110 can be configured to retrieve the component pixel data from the buffer 120 and to pass it to the CC/CL correction unit 130. Moreover, the buffer controller unit 110 can also receive the component pixel data of the current pixel, Yin(k) and Cin(k), and store it in the storage buffer 120.
According to an exemplary embodiment, component pixel data of the first previous pixel in the first frame, F1, and the second previous pixel in the second frame, F2, are used to determine whether cross-color and/or cross-luma is present for the current pixel in the current frame, Fin. The temporal difference between the current frame, Fin, and the first previous frame, F1, and between the first previous frame, F1, and the second previous frame, F2, is determined by the analog video standard implemented. For instance, in NTSC systems, the chroma subcarrier frequency is chosen so that its phase rotates by 180 degrees between successive scan lines. Because each video frame has an odd number of scan lines, e.g., 525 scan lines, the chroma subcarrier phase also rotates by 180 degrees between successive frames. Alternatively, in PAL systems, the chroma subcarrier frequency is chosen so that its phase rotates by substantially 90 degrees or 270 degrees between successive scan lines. Because each video frame has an odd number of scan lines, e.g., 625 scan lines, the chroma subcarrier phase also rotates by 90 degrees or 270 degrees between successive frames. Accordingly, in PAL systems, the chroma subcarrier phase rotates by 180 degrees between successive frames that are two frames apart.
When cross-color is present in an NTSC composite video signal, the luma signal is mistakenly decoded as the chroma signal. The phase rotation causes the erroneously decoded luma signal to oscillate between two complementary colors such as green and magenta, or blue and yellow at a rate of two frames per cycle. That is, the luma signal appears to be a spectral energy that oscillates between two complementary colors represented by chroma signals 180 degrees out of phase with one another. Similarly, when cross-luma is present, the chroma signal is mistakenly decoded as the luma signal. The phase rotation causes the erroneously decoded chroma signal to oscillate between two darker and brighter values at a rate of two frames per cycle. That is, the chroma signal appears to be a spectral energy that oscillates between two brightness levels represented by fluctuations in the luma signal 180 degrees out of phase with one another.
Alternatively, when cross-color is present in a PAL composite video signal, the phase rotation causes the erroneously decoded luma signal to oscillate between two complementary colors at a rate of four, as opposed to two, frames per cycle. Similarly, when cross-luma is present, the phase rotation causes the erroneously decoded chroma signal to oscillate between two darker and brighter values at rate of four frames per cycle.
Thus, in NTSC systems where the frame rate is approximately 30 Hz, the cross-color and/or cross-luma, if present, will oscillate at approximately 15 Hz. Accordingly, cross-color and/or cross-luma can be detected by analyzing the component pixel data in a detection window 300a that includes the current frame and at least two previous frames immediately preceding the current frame (see
Alternatively, in PAL systems where the frame rate is approximately 25 Hz, the cross-color and/or cross-luma, if present, will oscillate at approximately 6.25 Hz. Thus, cross-color and/or cross-luma can be detected by analyzing the component pixel data in a detection window 300b that includes the current frame, and at least two previous frames where the first of the two previous frames precedes the current frame by two frames and the second previous frame precedes the current frame by four frames (see
Referring again to
As stated above, the chroma pixel data, C, represents two color difference components, e.g., Cb pixel data and Cr pixel data. In this discussion, the cross-color determination module 410 can be either a Cb cross-color determination sub-module that receives and processes the Cb pixel data of the chroma pixel data, or a Cr cross-color determination sub-module that receives and processes the Cr pixel data. In either case, unless otherwise noted, the operation of the Cb and Cr cross-color determination sub-modules is identical to that of the cross-color determination module 410. Accordingly, for the sake of clarity, the cross-color determination module 410 will be described generally in relation to chroma pixel data, C, with an understanding that the chroma pixel data, C, can be either Cb pixel data or Cr pixel data.
According to one embodiment, the cross-color determination module 410 calculates a first chroma pixel data difference, ΔC1, by subtracting the chroma pixel data of the first previous pixel, C1, from the chroma pixel data of the current pixel, Cin, a second chroma pixel data difference, ΔC2, by subtracting the chroma pixel data of the second previous pixel, C2, from C1, and a third chroma pixel data difference, ΔC3, by subtracting C2 from Cin. Similarly, the cross-luma determination module 420 calculates a first luma pixel data difference, ΔY1, by subtracting the luma pixel data of the first previous pixel, Y1, from the luma pixel data of the current pixel, Yin, a second luma pixel data difference, ΔY2, by subtracting the luma pixel data of the second previous pixel, Y2, from Y1, and a third luma pixel data difference, ΔY3, by subtracting Y2 from Yin. The cross-color determination module 410 and the cross-luma determination module 420 are configured, in one embodiment, to detect characteristic patterns of cross-color and cross luma, respectively, based on the chroma pixel data differences and on the luma pixel data differences, respectively.
For example, the chroma pixel data of the current, first previous and second previous pixels form a characteristic cross-color pattern when the chroma pixel data of the current, first previous and second previous pixels form a high-low-high pattern or a low-high-low pattern, as depicted in
ΔC1>0 and ΔC2<0; or ΔC1<0 and ΔC2>0.
Similarly, the luma pixel data of the current, first previous and second previous pixels form a characteristic cross-luma pattern when the luma pixel data of the current, first previous and second previous pixels form a high-low-high pattern or a low-high-low pattern. That is, the characteristic cross-luma pattern can be defined by one of two conditions:
ΔY1>0 and ΔY2<0; or ΔY1<0 and ΔY2>0.
In addition to detecting the characteristic patterns of cross-color and cross luma, the cross-color determination module 410 and the cross-luma determination module 420 impose at least one additional condition in order to reduce the likelihood of a false detection event based solely on the characteristic patterns of cross-color and cross luma. For example,
MAX[ABS(ΔC1),ABS(ΔC2)]≦CP1C (for cross-color)
MAX[ABS(ΔY1),ABS(ΔY2)]≦CP1Y (for cross-luma)
The first control parameters, CP1C and CP1Y, are predetermined values that determine the sensitivity of the detection unit 400. For instance, as the first control parameters decrease, fewer cross-color and cross-luma events will be detected and vice versa. In one embodiment, the cross-color first control parameter CP1C and the cross-luma first control parameter CP1Y can be identical. In other embodiments, the cross-color first control parameter CP1C can differ from the cross-luma first control parameter CP1Y to reflect viewer preferences.
According to another embodiment, additional detection conditions can be imposed to increase reliability and robustness. For example, for a cross-color detection event to be true, the following additional conditions can be imposed:
MAX{MIN[ABS(ΔC1),ABS(ΔC2)]/2CP2C,CP3C}≧ABS(ΔC3); and
ABS(ΔC3)≦CP4C
Similarly, for a cross-luma detection event to be true, the following additional conditions can be imposed:
MAX{MIN[ABS(ΔY1),ABS(ΔY2)]/2CP2Y,CP3Y}≧ABS(ΔY3); and
ABS(ΔY3)≦CP4Y
Other or additional detection conditions can be imposed to customize the detection process. By imposing the additional detection conditions, what qualifies as a cross-color or cross-luma detection event can be more narrowly defined such that detection errors are minimized. For example,
Referring again to
According to an exemplary embodiment, the CC/CL correction unit 130 includes means for determining the per pixel count associated with the component pixel data of the current pixel. In one embodiment, the cross-color determination module 410 and the cross-luma determination module 420 can be configured to perform this function. For example, when the cross-color determination module 410 determines that cross-color is present for the current pixel, the chroma per pixel count associated with the chroma pixel data of the current pixel is determined to be the chroma per pixel count associated with the chroma pixel data of the first previous pixel incremented by one (1). That is:
Ccnt(x,y,k)=Ccnt(x,y,k−p)+1
When the cross-color determination module 410 determines that cross-color is not present, the chroma per pixel count associated with the chroma pixel data of the current pixel is set to zero (0). Accordingly, the chroma per pixel count can indicate whether cross-color has been determined to be present and in how many consecutive frames cross-color has been determined to be present. In one embodiment, the cross-color determination module 410 passes the chroma per pixel count associated with the chroma pixel data of the current pixel to the buffer controller unit 110 (
Similarly, when the cross-luma determination module 420 determines that cross-luma is present, the luma per pixel count associated with the luma pixel data of the current pixel is determined to be the luma per pixel count associated with the luma pixel data of the first previous pixel incremented by one (1). That is:
Ycnt(x,y,k)=Ycnt(x,y,k−p)+1
When the cross-luma determination module 420 determines that cross-luma is not present, the luma per pixel count associated with the luma pixel data of the current pixel is set to zero (0). Accordingly, the luma per pixel count can indicate whether cross-luma has been determined to be present and in how many consecutive frames cross-luma has been determined to be present. In one embodiment, the cross-luma determination module 420 passes the luma per pixel count associated with the luma pixel data of the current pixel to the buffer controller unit 110 so that the luma per pixel count can be stored in the storage buffer 120 with the luma pixel data of the current pixel.
To summarize, according to the embodiments described, the cross-color determination module 410 receives and analyzes the chroma pixel data of the current pixel, Cin, and two previous pixels, C1 and C2, to determine whether cross-color is present for the current pixel, determines the chroma per pixel count for the current pixel, Ccnt(x, y, k), based on the cross-color determination, and outputs the chroma per pixel count. Similarly, the cross-luma determination module 420 receives and analyzes the luma pixel data of the current pixel, Yin, and two previous pixels, Y1 and Y2, to determine whether cross-luma is present, determines the luma per pixel count for the current pixel, Ycnt(x, y, k), based on the cross-luma determination, and outputs the luma per pixel count.
The first input value is the per pixel count associated with the component pixel data of the first previous pixel incremented by one (1), and the second input value is zero (0). Note that the first input value saturates at a maximum value, e.g., 15, in order to reduce the number of bits necessary to represent the per pixel count. When the value of the bit received by the multiplexer 422 is one (1), meaning the presence of cross-color or cross-luma has been determined, the multiplexer 422 outputs the first input value as the luma or chroma per pixel count associated with the luma or chroma pixel data of the current pixel. Alternatively, when the value of the bit received is zero (0), meaning the presence of cross-color or cross-luma has not been determined, the multiplexer 422 outputs zero (0) as the luma or chroma per pixel count associated with the luma or chroma pixel data of the current pixel.
Referring again to
Similar to the cross-color determination module 410, the cross-color suppression module 610 can be either a Cb cross-color suppression sub-module that receives and processes the Cb pixel data of the chroma pixel data, or a Cr cross-color suppression sub-module that receives and processes the Cr pixel data. In either case, unless otherwise noted, the operation of the Cb and Cr cross-color suppression sub-modules is identical to that of the cross-color suppression module 610. Accordingly, for the sake of clarity, the cross-color suppression module 610 will be described generally in relation to chroma pixel data, C, with an understanding that the chroma pixel data, C, can be either Cb pixel data or Cr pixel data.
According to one embodiment, the cross-color suppression module 610 and the cross-luma suppression module 620 suppress cross-color and cross-luma, respectively, using pixels in a suppression window 310a, 310b that includes the current pixel and the first previous pixel (see
Cout(x,y,k)=Cin(x,y,k)+WFC×[C1(x,y,k−p)−Cin(x,y,k)]/2,
where WFc is a cross-color weighting factor having a value of at least zero (0) and at most one (1).
In one embodiment, the cross-color weighting factor is determined by the value of the chroma per pixel count associated with the chroma pixel data of the current pixel, Ccnt(x, y, k). The cross-color weighting factor can be, in one embodiment, non-decreasing with respect to the value of the chroma per pixel count. For example, when the chroma per pixel count value is zero (0), meaning the presence of cross-color has not been determined, the value of the cross-color weighting factor can be zero (0). When the cross-color weighting factor is zero (0), frame averaging is negated, and Cout(x, y, k)=Cin(x, y, k). Alternatively, when the chroma per pixel count value reaches the maximum value, meaning the presence of cross-color has been determined for a maximum number of consecutive frames including the current frame, the value of the cross-color weighting factor can be one (1). When such is the case, full frame averaging is performed, and Cout(x, y, k)=[Cin(x, y, k)+C1(x, y, k−p)]/2.
Similarly, the cross-luma suppression module 620 receives the luma pixel data of the current pixel, Yin(x, y, k), and the luma pixel data of the first previous pixel, Y1(x, y, k−p), and performs weighted frame averaging between the received luma pixel data to suppress the cross-luma. For example, the modified luma pixel data of the current pixel can be calculated by the following equation:
Yout(x,y,k)=Yin(x,y,k)+WFY×[Y1(x,y,k−p)−Yin(x,y,k)]/2,
where WFY is a cross-luma weighting factor having a value of at least zero (0) and at most one (1).
In one embodiment, the cross-luma weighting factor is determined by the value of the luma per pixel count associated with the luma pixel data of the current pixel, Ycnt(x, y, k). The cross-luma weighting factor can be, in one embodiment, non-decreasing with respect to the value of the luma per pixel count. For example, when the luma per pixel count value is zero (0), meaning the presence of cross-luma has not been determined, the value of the cross-luma weighting factor can be zero (0), frame averaging is negated, and Yout(x, y, k)=Yin(x, y, k). Alternatively, when the luma per pixel count value reaches the maximum value, meaning the presence of cross-luma has been determined for a maximum number of consecutive frames including the current frame, the value of the cross-luma weighting factor can be one (1), full frame averaging is performed, and Yout(x, y, k)=[Yin(x, y, k)+Y1(x, y, k−p)]/2.
In one embodiment, the cross-color weighting factor is received by a multiplier 630, which also receives a difference between the chroma pixel data of the first previous pixel and the current pixel. The product is divided by two (2) and then added to the chroma pixel data of the current pixel. In one embodiment, the modified chroma pixel data, e.g., Cout(x, y, k), can be received by a multiplexer 632 that is controlled by a suppression enabling bit. When the enabling bit value is one (1), the multiplexer 632 outputs the modified chroma pixel data. Otherwise, when the enabling bit is zero (0), the multiplexer 632 outputs the original chroma pixel data of the current pixel.
Similarly, the cross-luma weighting factor is received by a multiplier 630, which also receives a difference between the luma pixel data of the first previous pixel and the current pixel. The product is divided by two (2) and then added to the luma pixel data of the current pixel. In one embodiment, the modified luma pixel data, e.g., Yout(x, y, k), can be received by a multiplexer 632 that is controlled by the suppression enabling bit. When the enabling bit value is one (1), the multiplexer 632 outputs the modified luma pixel data. Otherwise, when the enabling bit is zero (0), the multiplexer 632 outputs the original luma pixel data of the current pixel.
Referring again to
In the third scenario, the CC/CL correction unit 130 determines that cross-color or cross-luma is present, and the per pixel count incrementally increases from zero (0) to a maximum value, M. In this embodiment, weighted frame averaging begins after a delay of approximately five (5) consecutive frames to ensure that the fluctuations in the component pixel data are due to cross-color or cross-luma. Thereafter, as the per pixel count increases, the degree of frame averaging increases until full frame averaging is performed.
By averaging the component pixel data of the current pixel and the first previous pixel, the out-of-phase cross-color or cross-luma in the component video signal can effectively be cancelled, as shown in
The CC/CL correction unit 130, in one embodiment, can be a stand alone system 100 that includes the buffer controller unit 110 and the storage buffer 120. In this embodiment, the system 100 can be inserted between a traditional composite video signal decoder/demodulator and a traditional display device and so that the component pixel data outputted by the decoder/demodulator is intercepted by the system 100 and the outputted modified component pixel data, Fout(k), can be received and displayed substantially without swirling rainbows, crawling dots, and other artifacts due to cross-luma or cross-color. Because the CC/CL correction unit 130 is configured to process component pixel data, the decoder/demodulator can use simple notch filters or 2-D comb filters, which are relatively inexpensive and computationally simple.
In yet another embodiment, the CC/CL correction unit 130 can be incorporated into any system that performs video signal processing, such as an image capture device, a set-top box, or a television display system. For example, the CC/CL correction unit 130 can be integrated with a de-interlacer in a television display system, where the de-interlacer converts an interlaced component video signal into a progressive component video signal which is subsequently displayed to a viewer.
In one embodiment, the integrated CC/CL correction unit 130 receives the interlaced component video input signal 29, Fin(k), automatically detects and suppresses cross-color and cross-luma artifacts present in the input signal 29, Fin(k), and outputs the modified component pixel data, Fout(k), to the noise reduction unit 910, which digitally reduces the noise present therein. The motion detection unit 920 receives the noise-filtered pixel data from the noise reduction unit 910 and calculates motion data related to a missing target pixel in an interlaced video field for which a value must be determined in order to generate the corresponding progressive component video signal. The motion data is received by the video source identification unit 930, which identifies the film mode of the video source based on the motion data. The video processing unit 940 receives the motion data from the motion detection unit 920 and a film mode indicator from the video source identification unit 930 and generates the progressive component video output signal 32, Pout(k).
In this embodiment, the CC/CL correction unit 130 shares pixel data read and write modules (not shown), the buffer controller unit 110 and the external storage buffer 120 with the motion adaptive de-interlacer 900. By sharing resources that are already available, the CC/CL correction unit 130 can be implemented without unduly impacting the complexity and cost of the overall display system 20.
According to the embodiments described, the CC/CL correction unit 130 can automatically detect and suppress cross-color and cross-luma artifacts from the pixel data of a component video signal derived from a composite video signal with a quadrature-modulated chroma signal mixed with a baseband luma signal. The resulting modified component video signal can then be subsequently processed and displayed substantially without swirling rainbows, crawling dots, and other artifacts due to crosstalk between the chroma and luma channels in a composite video signal. When implemented in a display system, such as a television, the CC/CL correction unit 130 can achieve video quality close to that of a system that uses 3-D comb filters without the expense and complexity associated with such filters.
The CC/CL correction unit 130 offers several significant advantages over existing cross-color/cross-luma suppression systems. For example, the cross-luma and cross-color detection and suppression can be performed individually or simultaneously in parallel with matched delay. Moreover, the control parameters utilized by the detection unit 400 are fully programmable and can be assigned separately to achieve different behaviors for the different perception of the human eyes for cross-luma and cross-color artifacts. For instance, the detection and suppression of cross-color can be assigned to a more aggressive setting, i.e., entering into frame averaging earlier, to avoid rainbow artifacts at the expense of more color blurring, while the detection and suppression of cross-luma can be assigned to a more conservative setting, i.e., entering into frame averaging later, to avoid saw-tooth and motion blur artifacts at the expense of more crawling dots.
In addition, the CC/CL correction unit 130 performs detection and suppression only in the temporal domain, unlike most existing systems that perform filtering in the horizontal and vertical domain as well. By filtering in the temporal domain only, cross-luma and cross-color detection and suppression are independent of image scaling in the horizontal and vertical domains and de-interlacing. That is, the exemplary process can be applicable to either spatially enlarged or shrunk input video signals, and either interlaced or progressive input video signals. Moreover, because the CC/CL correction unit 130 processes baseband component video signals, it can process decoded and demodulated composite video signals, such as NTSC and PAL, as well as component video signals, such as YPbPr and YCbCr, which are originated from a composite video signal.
According to additional aspects, the CC/CL correction unit 130 can process baseband component video signals with or without chroma sub-sampling, i.e., sampling rate down conversion for the chroma signals with respect to the luma signal. For example, for YCbCr component video signals with 4:4:4 sampling format, one set of cross-luma determination 420 and suppression 620 modules is needed for the luma component Y and two sets of cross-color determination 410 and suppression 610 modules are needed for the chroma components Cb and Cr, respectively. For YCbCr component video signals with 4:2:2 sampling format, one set of cross-luma determination 420 and suppression 620 modules is needed for the luma component Y and only one set of cross-color determination 410 and suppression 610 modules is needed for both chroma components Cb and Cr, which operates in a time multiplexing mode between Cb and Cr.
In another aspect, the CC/CL correction unit 130 does not rely on detecting motion in the luma and/or chroma component signals, unlike many other existing systems that measure motion in the luma signal to detect the stillness of the input video. Instead, the CC/CL correction unit 130 detects cross-luma and cross-color directly based on their characteristic frequencies. By not relying on motion detection, erroneous cross-color and cross-luma detection events can be minimized and improper filtering can be avoided.
While memory sharing with other video processing blocks in the video signal path, such as de-interlacing, is possible and desirable, another additional benefit is that memory saving can be achieves by employing the method and system described herein. By sharing the storage buffer 120, whose size is often determined by the highest definition signal formats that the system needs to support, there is actually no additional size for the storage buffer 120 required to store the per pixel count data and the component pixel data for the CC/CL correction unit 130. This is because the size for the storage buffer 120 is often determined by the required frame buffer size for the de-interlacing and/or frame-rate conversion of the highest definition signal formats, such as 1080i or 1080p, supported by the system, without considering the need for cross-luma and cross-color detection and suppression for such high quality high-definition (HD) signals, which generally are not originated from a composite video signal. While for those lower quality standard-definition (SD) signals, such as 480i for NTSC or 576i for PAL, where cross-luma and cross-color detection and suppression are much more desired, the required frame buffer size for the de-interlacing and frame-rate conversion is much less. Hence, the CC/CL correction unit 130 utilize the spare memory space that already exists in the storage buffer 120 for the support of HD signals. Accordingly, additional space for the storage buffer 120 is not required to employ the method and system described herein for SD signals.
According to embodiments described, the cross-color and cross-luma of a baseband component video signal originated from a quadrature-modulated composite video signal can be detected and suppressed, and hence the output baseband component video signal can be correctly processed and displayed substantially without swirling rainbows, crawling dots, and other artifacts due to cross-color and cross-luma. It should be understood that the various components illustrated in the figures represent logical components that are configured to perform the functionality described herein and may be implemented in software, hardware, or a combination of the two. Moreover, some or all of these logical components may be combined and some may be omitted altogether while still achieving the functionality described herein.
It will be understood that various details of the invention may be changed without departing from the scope of the claimed subject matter. Furthermore, the foregoing description is for the purpose of illustration only, and not for the purpose of limitation, as the scope of protection sought is defined by the claims as set forth hereinafter together with any equivalents thereof entitled to.
| Number | Name | Date | Kind |
|---|---|---|---|
| 4706112 | Faroudja et al. | Nov 1987 | A |
| 4731660 | Faroudja et al. | Mar 1988 | A |
| 5305120 | Faroudja | Apr 1994 | A |
| 5502509 | Kurashita et al. | Mar 1996 | A |
| 7271850 | Chao | Sep 2007 | B2 |
| 7460180 | Chao | Dec 2008 | B2 |
| 7697075 | Zhu | Apr 2010 | B2 |
| 20050157789 | Chao | Jul 2005 | A1 |
| 20050280740 | Chao | Dec 2005 | A1 |
| 20060017854 | Chao | Jan 2006 | A1 |
| 20060092331 | Zhu | May 2006 | A1 |
| 20060092332 | Zhu | May 2006 | A1 |
| 20060152631 | Chang | Jul 2006 | A1 |
| 20060176405 | Chen | Aug 2006 | A1 |
| 20060181648 | Park et al. | Aug 2006 | A1 |
| 20070024761 | Chang | Feb 2007 | A1 |
