JP5045119B2 - Color transient correction device - Google Patents

Description

  The present invention relates to a color transient correction apparatus used in video equipment such as a television, a projector, a video, and a digital camera that performs digital video signal processing.

  In recent years, digital processing of video signals has become common in video equipment such as televisions. Such a digital video signal is sometimes expressed as luminance (Y) and color difference (Cb, Cr or Pb, Pr, but is described here as U and V. Strictly speaking, the matrix coefficients are different, but It is not related to the purpose, so it is described as equivalent.) In addition, a clock and a synchronizing signal (H, V) are necessary on the circuit, but this is well known and will not be described in detail.

  On the other hand, in terms of human visual characteristics, color components are not noticeable even if the resolution is lower than the luminance component, and are often handled with the band of the color difference signal reduced. This can be used to reduce the clock rate of the color difference signal, thereby reducing the circuit scale and power consumption.

  For example, in general, when 4: 4: 4 is expressed among engineers in the field, luminance signals and color difference signals (U, V) are sampled at the same frequency, but expressed as 4: 2: 2. This means that the color difference signals (U, V) are sampled at half the frequency of the luminance signal (1/2 sampling rate). 4: 1: 1 means that the color difference signal has a sampling rate of 1/4 of the luminance signal. In 4: 2: 0, after 4: 2: 2, the vertical direction is also thinned by half.

  FIG. 5 shows a signal format at 4: 4: 4 and a signal format at 4: 2: 2. Thus, in 4: 2: 2, the color difference signal can be reduced from two to one.

  FIG. 6 shows an example of a block diagram when converting (decoding) from 4: 2: 2 to 4: 4: 4.

  In FIG. 6, FF1 is a flip-flop, and although not shown, it functions to delay the input signal by a clock signal. LH2 is a load / hold, and the input signal is delayed by a clock signal (not shown) as in the case of FF1, only when the UV selection pulse is in the H period. The same applies to LH3. The output is not changed when the UV selection pulse is in the L period. Both FF and LH are basic elements of digital circuits and are well known among digital engineers.

  As described above, the multiplexed color difference signal is delayed by FF (flip flop) and taken out by LH (load hold), so that decoding to 4: 4: 4 becomes possible. FIG. 7 shows a timing chart. A signal delayed by an FF or LH delay circuit in the circuit is output, but delay adjustment is usually performed by delaying other signals such as a synchronization signal and a luminance signal in accordance with this delay. Such a delay adjustment is well known and is omitted in the figure.

  By the way, as shown in FIG. 7, if the color signal remains unchanged, the same value continues every two pixels for even and odd color signals. For example, the output of LH2 is V0, V0, V2, and V2. This is because in the case of 4: 4: 4 format as shown in FIG. 5, when signals V0, V1, V2, and V3 are output, odd-numbered signals such as V1 and V3 are deleted at the time of input. As a result, when the decoded signal is viewed as an image, it becomes an unnatural image with a color change and the like rattling. Therefore, in the conventional color transient correction apparatus, an odd-numbered pixel is interpolated from the previous and subsequent pixels (for example, see Patent Document 1). A typical one used for the interpolation is the following linear interpolation (Equation 1).

    However, n is n = 0, 1, 2, 3,. In this way, it has been improved that an unnatural image is obtained by adding the previous and subsequent pixels and dividing by two.

However, there is a problem that the color band remains limited and color blurring occurs. In the configuration cited in Patent Document 2 as a countermeasure, each of the luminance signal and the color difference signal is band-divided by wavelet transform, and the high-frequency color difference signal is estimated from the high-frequency luminance signal based on the correlation between the edge positions of both signals. An enhanced color difference signal is output.
Japanese Patent Laid-Open No. 2003-8865 JP-A-10-224815

  However, the latter method requires subband division such as wavelet transform, which increases the circuit scale. As described above, the former method has a problem in that the luminance component changes at the transition point (transient). On the other hand, there is a problem that the color change point is blurred.

  In addition, when multi-screen processing such as two screens synthesized during 4: 2: 2 processing or on-screen display (OSD) processing for inserting characters or the like into the screen is added, the boundary between those processing is added. There is also a problem that colors are lost or colors are popped out.

  SUMMARY OF THE INVENTION The present invention solves the above-described conventional problems, and an object thereof is to provide a color transient correction apparatus that has a relatively small circuit scale and can be realized with a simple configuration and has improved color bleeding.

  In order to solve the above-described conventional problems, a color transient correction apparatus according to the present invention includes a UV selection pulse generation unit that outputs a UV selection pulse from an input synchronization signal, and a time-division multiplexed input color difference signal as the UV selection pulse. A color difference signal separating means for separating the first intermediate color difference signal and the second intermediate color difference signal based on the weight coefficient calculating means for calculating and outputting a weight coefficient from the input luminance signal; and the first color difference signal based on the weight coefficient. A first interpolation means for interpolating the intermediate color difference signal and outputting a first output color difference signal; a second interpolation means for interpolating the second intermediate color difference signal based on the weighting factor and outputting a second output color difference signal; Interpolating means.

  With this configuration, since the coefficient of the interpolation filter is changed according to the change in the luminance signal, it is possible to provide a color transient correction apparatus that does not cause color blurring with a simple configuration.

According to the color transient correction apparatus of the present invention, it is possible to provide a color transient correction apparatus with a relatively small circuit scale and an improved color blur, and the composition can be performed during 4: 2: 2 processing. Even if on-screen display (OSD) processing that inserts multi-screen processing such as two screens and characters etc. into the screen is added, colors may be lost or jump out at the boundary of those processing No color transient correction device can be provided.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

(Embodiment 1)
FIG. 1 is a block diagram of a color transient correction apparatus according to Embodiment 1 of the present invention.

  FIG. 2 is a timing chart showing the operation of the color transient correction apparatus according to this embodiment.

  In FIG. 1, the input signal indicates a signal of 4: 2: 2 format, and its timing chart is like “input luminance signal Y”, “input color difference signal C”, “input synchronization signal” shown in FIG. It has become. The color difference signal separation means 101 is a circuit that extracts U and V signals multiplexed on the “input color difference signal C” and outputs them as “color difference signal U” and “color difference signal V” in FIG. As a specific circuit example, a combination of FF and LH shown in FIG. 6 is common.

  The UV selection pulse generation means 102 receives a synchronization signal and outputs a UV selection pulse to the color difference signal separation means 101 described above. As a specific circuit example, it is well known to digital engineers in the field, so detailed description using a circuit diagram is omitted, but a clock signal corresponding to a video signal (omitted in the block diagram) is input as a clock. 1 is added to the output of the FF, and 1 is added to the output of the FF, and the result is passed through the selector so as to be reset by a synchronization signal, and then input to the FF described above to make a basic circuit. It is possible. However, it is necessary to adjust the delay according to the actual circuit delay. In the case of creation by software, the color difference signal is usually expressed by an array, and can be determined by referring to the subscript of the array and determining whether it is an even number or an odd number.

  The interpolating means 103 converts the signals of V1, V3, V5, V2n + 1 (where n is n = 0, 1, 2,...) Of the “output color difference signal V” in FIG. ) To obtain by interpolation. The following (Formula 2) is used as an example for the interpolation formula.

    α indicates an interpolation phase and is used in a range of α = 0 to 1.0. When α is 0, a value close to V2n + 2 is obtained as an interpolation value, and when α is 1, the value is close to V2n. As described above, one of the features of the present embodiment is to use an interpolation filter whose phase characteristics are different depending on α. Therefore, although a general linear interpolation equation is shown as an example of the interpolation equation, it is needless to say that another interpolation function may be used. In addition, although α is expressed as a real number instead of an integer, in the actual circuit or arithmetic processing, a finite digit operation using a fixed point, that is, a multiple of 2 is multiplied by α to make an integer, and an integer operation is performed. Needless to say, the process of dividing the number by a multiple of 2 (truncating the lower bits) is sufficient in terms of speed, circuit scale, and the like.

  Note that α is obtained from the weighting factor calculation means 105 described later.

  The interpolating means 104 is similar to the interpolating means 103, and the signal handled only changes from V to U.

  On the other hand, the weight coefficient calculation means 105 is means for calculating α in accordance with the change in luminance before and after Y2n + 1 corresponding to V2n + 1 and U2n + 1. For example, the following (Formula 3) is performed as an example.

    However, max (a, b) indicates the larger of a and b, and min (a, b) indicates the smaller of a and b.

  FIG. 3 shows the relationship between the luminance signal subscripts and the color difference signal subscripts shown in the equation. In FIG. 3, the horizontal direction indicates the horizontal position of the signal, and the vertical direction indicates the value of each luminance signal and color difference signal using ◯. Note that ● of the color difference signal originally indicates a portion having a value in 4: 2: 2, and in this example, an even-numbered signal is indicated. The ◯ of the color difference signal indicates a portion created by interpolation. In this example, an odd-numbered signal is indicated. In the figure, ◯ indicates the state output by the color difference signal separating means 101, and thus has the same value as the previous color difference signal.

  The meaning of the above formula is to check whether the luminance at the target pixel Y2n + 1 is correlated with the previous pixel Y2n or the subsequent pixel Y2n + 2, and if the previous pixel is correlated, α is increased. If there is a correlation in the subsequent pixels, α is made smaller. In addition, when the pixel of interest Y2n + 1 is not between the preceding and succeeding pixels (Y2n, Y2n + 2), a process of clipping within the range is also included.

  For example, when the if clause in the previous equation is true, the previous pixel Y2n is large and the subsequent pixel Y2n + 2 is small. Therefore, the maximum value of the pixel of interest Y2n + 1 must be Y2n. That is, the smaller one of Y2n + 1 and Y2n may be selected. The portion showing the processing corresponds to min (Y2n, Y2n + 1). Further, since the value is at least Y2n + 2 or more, it is only necessary to output the larger one of Y2n + 2, so the expression is max (Y2n + 2, (min (Y2n, Y2n + 1)). From the result, the distance from Y2n + 2 can be obtained by subtracting Y2n + 2, and the correlation can be obtained by dividing it by (Y2n−Y2n + 2). A value of 1 indicates that the previous pixel (Y2n) has the strongest correlation.

  Although the correlation detection method is shown by such an expression, it is of course possible to detect the correlation by other methods. For example, not only from two points before and after, but also from a plurality of points, a multi-order curve is obtained and the correlation from the curve is also obtained. It goes without saying that the correlation from the curved surface may be obtained by extending to two dimensions.

  Interpolation is performed by using the correlation result thus obtained as α as the interpolation phase of the interpolation means 103 and the interpolation means 104 as described above.

  FIG. 4 shows an outline of the operation of interpolation of the intermediate color difference signal in the present embodiment. Three examples (a), (b), and (c) are shown. The horizontal position of the pixel is shown horizontally, the value of each signal is shown vertically, the upper part shows the luminance signal, and the lower part shows the color difference signal. In the color difference signal, a signal in which black originally exists (even-numbered signal) and a signal in which white is interpolated (odd-numbered signal) are shown. The center is the point of interest.

  In (a) in the figure, the luminance signal at the point of interest protrudes below the previous and subsequent signals, and the above-described clip processing is performed. The signal after clip processing is indicated by a dotted line. Since α is obtained from this dotted signal, α is 1.0. On the other hand, since the color difference signal is α = 1, it is interpolated using the previous signal and is interpolated as shown by the arrows in the figure.

  In (b) in the figure, the luminance signal is in the middle of the preceding and following signals, and is 80% closer to the front luminance signal, so α is 0.8. Therefore, the color difference signal is interpolated at the position shown in the figure by linear interpolation between the front side and the rear side.

  In (c) in the figure, the luminance signal is clipped upward, and α = 0. Therefore, the color difference signal is interpolated to the same value as the rear signal.

  As can be seen from the above description, the color signal interpolation phase is obtained from the correlation from the surrounding pixels of the luminance signal corresponding to the color signal interpolation point, and is actually used for color signal interpolation to improve color transients. It becomes possible to make it.

  In addition, an example in the case of converting from 4: 2: 2 to 4: 4: 4 (YUV) is shown, but the conversion from 4: 1: 1: 4 to 4: 4: 4 is also a correlation between the color difference interpolation formula and the luminance. Needless to say, the concept is the same and is not limited to the above example.

  It goes without saying that the 4: 2: 0 format, which is thinned out in the vertical direction, can be handled by applying the vertical application in the same manner.

  In addition, the direction of α is set to 0 when the correlation is high in the subsequent pixels, but it is needless to say that the reverse is also possible.

(Embodiment 2)
In the present embodiment, in addition to the configuration and operation of the first embodiment, when the denominator is smaller than a predetermined threshold, α is calculated so as to approach the midpoint α = 0.5.

  For example, when the denominator is smaller than the predetermined threshold β, the calculation is performed according to the following (Equation 4).

    However, max (a, b) indicates the larger of a and b, and min (a, b) indicates the smaller of a and b.

  Thereby, when there is a slight change due to noise on the luminance signal in a portion where the luminance change is small and the color change is large, it is possible to prevent the noise from doubling in the color direction.

  As described above, even when the luminance signal fluctuates slightly due to noise or the like, the influence on the color can be reduced.

(Embodiment 3)
In the weight coefficient calculating means 105 having the above-described configuration, division is used for calculating α, and there is a problem that the calculation speed becomes slow because the circuit scale has increased. In this embodiment, the input is D and M, and a two-dimensional look-up table (LUT) that returns α as an output. However, D is D = | Y2n−Y2n + 2 |, and M is expressed by the following (Equation 5).

    The look-up table is a table in which values corresponding to the values of D and M are stored in advance in a ROM (Read Only Memory) or the like and read according to D and M.

  By making the value to be written in this look-up table the result of calculation using the formulas described in the first and second embodiments, problems of circuit scale and calculation speed can be avoided.

  Needless to say, any value can be entered without being bound to the above expression.

  Since the color transient correction apparatus of the present invention can reduce color blur and correct color misregistration with a simple configuration, it can be applied to all digital devices that perform digital video signal processing, such as televisions, videos, DVD players and recorders, It is also useful for digital still cameras, video cameras, and software for image processing.

1 is a block diagram showing the configuration of a color transient correction apparatus according to Embodiment 1 of the present invention. FIG. 3 is a timing chart showing the operation of the color transient correction apparatus according to Embodiment 1 of the present invention. The figure which shows a response | compatibility of the luminance signal and color difference signal in Embodiment 1 of this invention. The figure explaining the interpolation phase in Embodiment 1 of this invention Diagram showing 4: 4: 4 and 4: 2: 2 format Circuit diagram of color difference signal separating means 101 Timing chart of color difference signal separating means 101

Explanation of symbols

1 FF (flip flop)
2, 3 LH (load hold)
101 Color difference signal separating means 102 UV selection pulse generating means 103, 104 Interpolating means 105 Weight coefficient calculating means

Claims (1)

UV selection pulse generating means for outputting a UV selection pulse from the input synchronization signal;
Color difference signal separation means for separating the time-division multiplexed input color difference signal into a first intermediate color difference signal and a second intermediate color difference signal based on the UV selection pulse;
Means for calculating and outputting a weighting factor from the input luminance signal, wherein when the change of the input luminance signal is smaller than a predetermined threshold, a weighting factor calculating unit for calculating the weighting factor to approach 0.5 ;
First interpolation means for interpolating the first intermediate color difference signal based on the weighting factor and outputting a first output color difference signal;
A color transient correction apparatus comprising: a second interpolation unit that interpolates the second intermediate color difference signal based on the weighting factor and outputs a second output color difference signal.

Images