KR20030058194A - Filter for eliminating a dot pattern and method thereof - Google Patents

Abstract

The present invention relates to a dot pattern removal filter and a method thereof. In particular, a dot appearing at a boundary of an NTSC color TV signal demodulated by incomplete luminance / color signal separation by removing a luminance / color signal interference component from a decoded component video signal region. ) To remove the pattern. An object of the present invention is to provide an NTSC decoder which demodulates a component video signal (YIQ) by NTSC decoding by removing a dot pattern included in an NTSC composite video signal (CV), and a component video signal demodulated by the NTSC decoder. By adjusting the weights according to the transition region detection of the chrominance signal and the dot pattern detection of the luminance, the cutoff frequency for the low pass filtering in the vertical / horizontal direction is adjusted and the detection of the transition region of the luminance and the dot pattern detection of the chrominance signal It is characterized in that it comprises a noise canceller for removing the dot pattern appearing in the luminance signal and the chrominance signal by adjusting the cutoff frequency for the low pass filtering in the vertical / horizontal direction by adjusting the weight.

Description

FILTER FOR ELIMINATING A DOT PATTERN AND METHOD THEREOF}

TECHNICAL FIELD The present invention relates to image processing technology, and more particularly, to a dot pattern removal filter included in a demodulated TV signal and a method thereof.

In general, video signals acquired by a camera and transmitted through a channel are essentially noise added during video acquisition and transmission.

Such noise improperly affects the video signal processing to provide a degraded image to the viewer and to provide a higher quality image.

Noise also increases the entropy of the video signal, thereby reducing the video compression efficiency.

Therefore, in the pre-processing of the video compression encoder or the post-processing of the decoder, a noise removing process for solving such a problem is performed.

In particular, in addition to general noise, color TV signals such as NTSC, which is the current domestic TV standard, generate noise included in the color TV signals.

The color TV signal of the NTSC method is obtained by interleaving a color signal that is orthogonally modulated with a color subcarrier frequency of about 3.58 MHz.

Therefore, when the separation between the luminance signal and the color signal is incomplete, a cross-color phenomenon in which part of the luminance signal is included in the color signal and a cross-luminance phenomenon in which part of the color signal is included in the luminance signal are generated.

Since these phenomena appear as rainbow patterns and dot patterns in the reproduced image, the degradation of image quality also reduces the video compression efficiency.

In the general NTSC decoding circuit, as shown in the block diagram of FIG. 1, the NTSC decoder 110 demodulates an NTSC composite video signal (CV: Composite Video) into a component video in YIQ or YUV format by NTSC decoding and performs digital video. The application block 120 inputs the component image and performs processing for digital image application.

The digital image application block 120 may apply digital image processing, multimedia processing, desktop video, video phone, and the like.

2 shows an NTSC encoding and decoding process.

2 (a) is a schematic diagram of an NTSC encoder, and as shown therein, a matrix circuit 211, a low pass filter 222, 223, 228, a mixer 224, 225, and an adder 226 ( 227).

The operation of the NTSC encoder is as follows.

In the case of encoding, an RGB signal obtained from a camera is converted into a luminance signal Y and a color difference signal I, Q by a matrix circuit 211, and the color difference signals I, Q are each low pass filter 212 ( Low pass filtering at 213 and then input to mixers 214 and 215.

At this time, the mixer 214 outputs the output signal from the low pass filter 212. And the mixer 215 converts the output signal from the low pass filter 213. Quadrature Amplitude Modulation (QAM) is performed at the color subcarrier frequency of about 3.58 MHz.

Thereafter, when the adder 216 sums the output signals of the mixers 214 and 215 and the adder 217 sums the output signals of the adder 216 to the output signal Y of the matrix circuit 211, the low pass. The filter 218 low-pass filters the composite video signal ( ) Will be printed.

The NTSC coded composite video signal CV is expressed as in Equation (1) below.

--------- (One)

here, Is the color carrier frequency Is the phase of the chrominance signal. to be.

FIG. 2 (b) is a schematic diagram of an NTSC decoder, as shown therein, Y / C separation circuit 221, mixer 222, 223, low pass filter 224, 225, and dematrix circuit. 226.

The operation of the NTSC decoder is as follows.

Decoding is performed in the reverse order of the encoding process.

The received composite video signal CV is separated into a luminance Y 'and a color signal C' in the Y / C separation circuit 221, and the color signal is mixed in the mixer 222. Is multiplied and at mixer 223 By multiplying it is demodulated back to the color carrier frequency.

Thereafter, when the low pass filter 224, 225 low pass filters the output signals of the mixers 222, 223, and converts them into the color difference signal I ″, Q ″, the dematrix circuit 226 performs Y / Y. The luminance signal Y 'in the C separation circuit 221 and the color difference signal I "and Q" in the low pass filter 224 and 225 are converted into an R'G'C' signal.

However, in the above decoding process, when the separation between the luminance and the color signal is incomplete, an interference phenomenon occurs in which the high frequency component of the luminance signal is included in the color signal and the high frequency component of the color signal is included in the luminance signal.

If the luminance and color signals are separated from the received luminance signal and the chrominance signal, the signal components (low frequency components) normally divided into luminance and color signals are respectively and And the luminance signal included in the color signal and the color signal component (high frequency component) included in the luminance signal, respectively. Wow In this case, the luminance / color signal separated luminance Y and the color signal C are expressed as in Equation (2) below.

-------- (2)

here, and Is a color difference signal component normally separated into color signals. Wow Is the chrominance signal component included in the luminance signal. to be.

As a result, the interference component of the color signal included in the luminance among the demodulated NTSC signal components is obtained from Equation (2). It can be seen that.

Since the interference component includes a sinusoidal component corresponding to the color subcarrier frequency, a dot pattern is generated in the luminance signal.

In addition, the color signal C ′ obtained in the luminance / color signal separation process is converted into the color difference signals I ″ and Q ″ by demodulation processing, which is an inverse process of orthogonal modulation of the encoder, which is expressed as in Equation (3) below.

-(3)

here, Is a lowpass filter with an output gain of '2'.

Among the NTSC signal components demodulated from Equation (3), interference components of the luminance signal included in the color difference signal (I) (Q) are respectively Wow It can be seen that.

This interference component appears in the form of a rainbow pattern when the amplitude of the actual color difference signal is small and in the form of a dot pattern when the amplitude is large.

On the other hand, the color carrier frequency ( ) And horizontal scan frequency ( ) Is shown in Equation (4) below.

---------------- (4)

Therefore, when the image signal input to the NTSC encoder is a flat image with no change, the color signal is in the horizontal direction, as shown in the example of FIG. The phase is reversed every cycle and every scan line.

In particular, in case of decoding by digital processing, If set to, the phase of the color signal is inverted every two samples in the horizontal direction.

Therefore, in the luminance / color signal separation process performed by the NTSC receiver, the luminance signal is easily extracted by low pass filtering in the case of a flat image region. The color signal is easily extracted by the band pass filter centered on the.

However, in the case of an image region having a sudden change, an interference phenomenon between the luminance and the color signal occurs because high frequency components of the luminance and the color signal exist.

That is, when the luminance / color signal separation performed by the NTSC decoder is incomplete, a dot pattern occurs during the luminance / color signal interference in a transition region in which the luminance or color signal change is large.

In other words, a dot pattern appearing in the luminance signal occurs when the color signal has a sudden change in the horizontal or vertical direction. This is because a quadrature amplitude modulated (QAM) color difference signal is incorporated into the luminance, so that the luminance pattern has a dot pattern unique to the NTSC signal. It happens.

In the dot pattern appearing in the demodulated color difference signal, the luminance signal is generated in the horizontal or vertical boundary region. Since the luminance signal mixed in the color signal is QAM demodulated, the dot pattern unique to the NTSC signal in the demodulated color difference signal is slightly reduced. Rainbow pattern will appear.

On the other hand, many studies have been conducted to suppress NTSC luminance / color signal interference phenomenon. Among the previous studies, the method of preventing the interference from being generated by performing the adaptive luminance / color signal separation in the NTSC decoder itself and the luminance / color signal separation There is a method of removing the luminance signal included in the color signal C '.

However, in the conventional methods, the interference component included in the demodulated component video cannot be removed because the interference component is removed in the undecoded composite video signal (CV) region.

In fact, most video-related application fields that use images acquired with conventional NTSC TV cameras use a specific NTSC decoder chip to perform the necessary processing as a component video signal, which is a decoded luminance and chrominance signal, as a composite video signal (CV). There is a need for a method of handling interference of luminance and color signals in a component video region, rather than a region).

Accordingly, the present invention eliminates the dot pattern created to remove the dot pattern appearing at the boundary of the NTSC color TV signal demodulated by incomplete luminance / color signal separation by removing the luminance / color signal interference component in the decoded component video signal region. An object of the present invention is to present a filter and a method thereof.

1 is a block diagram of a general NTSC decoding circuit.

2 is a block diagram showing an NTSC decoding and encoding process;

3 is an exemplary diagram showing a phase relationship of NTSC color signals.

Figure 4 is a block diagram of a dot pattern removal filter for an embodiment of the present invention.

Figure 5 is an exemplary view showing the shape of the horizontal, vertical dot pattern.

6 is a block diagram showing the configuration of a noise canceller in FIG.

7 is an exemplary view showing the operation of the transition region detector in FIG.

FIG. 8 is an exemplary view showing the operation of a dot pattern detector in FIG. 6; FIG.

9 is an exemplary view showing an operation concept of a low pass filter of FIG.

FIG. 10 is a circuit diagram showing the configuration of a low pass filter in FIG. 6; FIG.

FIG. 11 is a block diagram illustrating an embodiment of a weight generator in FIG. 6; FIG.

FIG. 12 is a block diagram showing another embodiment of a weight generator in FIG. 6; FIG.

FIG. 13 is a block diagram illustrating still another embodiment of the weight generator in FIG. 6; FIG.

14 is an exemplary view showing a truth table according to the operation of each part in FIG.

Explanation of symbols on the main parts of the drawings

410: NTSC decoder 420: digital video application block

430: noise canceller 611,612: line memory

621,631: dot pattern detector 622,632: transition region detector

623,633: weight generator 624,634: vertical low pass filter

625,635 horizontal low pass filter

The present invention provides an NTSC decoder for demodulating a component video signal (YIQ) by NTSC decoding by removing a dot pattern included in an NTSC composite video signal (CV) to achieve the above object, and demodulating in the NTSC decoder. And a noise canceller for removing interference noise included in the component image.

The noise canceller includes a line memory for temporarily storing a luminance / color difference signal, a vertical and horizontal low pass filter for passing only a predetermined low frequency component in the vertical and horizontal directions with respect to each of the luminance / color difference signals delayed by two lines in the line memory; A dot pattern detector for receiving a luminance / color difference signal from a line memory and detecting a horizontal and vertical dot pattern for each luminance / color difference signal for each pixel, and a horizontal and vertical signal receiving a color difference / luminance signal from the line memory. A weighted signal for controlling the cutoff frequency of the vertical and horizontal low pass filters by inputting a transition region detector for detecting a transition region in a direction, a dot pattern in the dot pattern detector and transition region information in the transition region detector Image quality due to luminance / color signal interference of NTSC signal with vertical / horizontal weight generator It characterized in that the screen arranged to inhibit.

In addition, the present invention is to remove the dot pattern included in the composite video signal to separate the component signal by NTSC decoding to achieve the above object, and to detect the presence or absence of the transition region for the output component signal And detecting the presence or absence of a dot pattern with respect to the component signal outputted from the output signal, and performing vertical and horizontal low pass filtering on the component signal when the transition region and the dot pattern are detected. It is done.

In the performing of the low pass filtering, the weight is adjusted according to the transition region detection signal of the color difference signal and the dot pattern detection signal of luminance to adjust the cutoff frequency of the horizontal / vertical low pass filter connected in cascade. The color difference signal is controlled by reducing the dot pattern appearing in the luminance and adjusting the cutoff frequency of the horizontal / vertical low pass filter connected in series by adjusting the weight according to the transition region detection signal of the luminance and the dot pattern detection signal of the color difference signal. Characterized in that the process to reduce the dot pattern appearing in the.

The weight generation is characterized in that the IIR filter concept is applied in consideration of stability.

Hereinafter, the present invention will be described in detail with reference to the drawings.

In the embodiments of the present invention, a detailed description of well-known functions and configurations that may unnecessarily obscure the subject matter of the present invention will be omitted.

As shown in the block diagram of FIG. 4, the circuit for removing the interference noise included in the demodulated component image in the NTSC decoder 410 is input to the digital image application block 420. 430 is configured to suppress deterioration of image quality due to luminance / color signal interference of the NTSC signal.

Referring to Figures 4 to 14 the operation and effect of the embodiment of the present invention configured as described above are as follows.

Dot noise caused by interference of luminance or color signals is mainly generated in the transition region where the luminance or color signal change is large.

Dot noise appearing in the luminance signal is generated by mixing high frequency components of the color signal into the luminance signal in the transition region of the color signal, and dot noise appearing in the color difference signal is generated by mixing the high frequency component of the luminance signal in the transition region of the luminance signal in the color signal.

As described above, the dot noise caused by the luminance or the color signal interference is expressed as the color carrier frequency in the horizontal direction as described in Equation (2) and (3) above. Sinusoidal shape with period of) and in the vertical direction the horizontal scanning frequency ( 2 times Sinusoidal form with period

5 is an exemplary view showing the shape of the dope pattern occurring in the horizontal and vertical transition region.

As shown in Fig. 5, the sampling frequency is the color carrier frequency ( 4 times of) ), The dot pattern generated in the vertical transition region represents a sinusoidal wave form in which the period is 4 pixels in the horizontal direction, that is, the phase is inverted every 2 pixels, and the dot pattern generated in the horizontal transition region corresponds to every line in the vertical direction. The phase is reversed every time.

Therefore, in the case of luminance, the dot pattern removing method proposed by the present invention blocks the horizontal / vertical low pass filter connected in series by determination information on the presence or absence of a transition region for the color difference signal and the presence or absence of a dot pattern on the luminance signal. By adjusting the frequency, the dot pattern is reduced, and in the case of the color difference signal, the horizontal / vertical low pass filter connected in series according to the judgment information about the presence or absence of the transition region for the luminance signal and the dot pattern for the color difference signal is blocked. By adjusting the frequency, the dot pattern is reduced.

FIG. 6 is a block diagram of a noise canceller for two-dimensional dot pattern removal according to an embodiment of the present invention. As shown therein, a line memory 611 for delaying the luminance signal Y by two lines and a color difference signal ( A line memory 612 for delaying two lines of I / Q, and a vertical and horizontal low pass filter 624 and 625 for passing only predetermined low-pass components in the vertical and horizontal directions with respect to the luminance signal of the line memory 621. ), A dot pattern detector 621 for detecting a dot pattern in the horizontal and vertical directions with respect to the luminance signal by receiving the luminance signal from the line memory 611, and a color difference signal from the line memory 612. A transition region detector 622 for detecting a transition region in the horizontal and vertical directions, and a dot pattern in the dot pattern detector 621 and transition region information in the transition region detector 622 as input. Vertical and horizontal Vertical / horizontal weight generator 623 for outputting a weighted signal for controlling the cutoff frequency of the low pass filter 624 and 625, and a vertical and horizontal direction with respect to the color difference signal delayed by two lines in the line memory 612. Vertical and horizontal low pass filters 634 and 635 passing only predetermined low-pass components of the pixel and the line memory 612 receive color difference signals to detect horizontal and vertical dot patterns with respect to the color difference signals for each pixel. A dot pattern detector 631, a transition region detector 632 for detecting a horizontal and vertical transition region by receiving a luminance signal from the line memory 611, and a dot in the dot pattern detector 631. A vertical / horizontal weight generator 633 for outputting a weighted signal for controlling a cutoff frequency of the vertical and horizontal low pass filters by inputting a pattern and transition region information from the transition region detector 632. Constitute over.

That is, in Fig. 6, the luminance and chrominance signals Y and I / Q, which are the component input signals, are input to the two-line memories 611 and 612, respectively, and the outputs of the respective line memories 611 and 612 are equal to the luminance. The dot pattern detectors 621 and 631 and the transition region detectors 622 and 632 for the color difference signal are input to the horizontal and vertical transition regions and the horizontal and vertical dots for the luminance and the color difference signals for each pixel. Determine the presence of a pattern.

The determination information about the transition region and the dot pattern for each of the luminance and the color difference obtained by the determination result is input to the vertical / horizontal (V / H) weight generators 623 and 633 to receive the vertical low pass filter (V−). Weighting signal for controlling the cutoff frequency of the LPF (624) 634 and the horizontal low pass filter (H-LPF) (625) (635) , Is converted to).

Here, the weighted signal generation can be applied to the IIR filter concept in consideration of stability.

That is, in the present invention, the dot pattern removing process removes the dot noise appearing in the luminance, and adjusts the weight according to the transition region detection signal of the color difference signal and the dot pattern detection signal of the luminance to connect the horizontal / vertical low pass filter connected in series ( 624,625) to reduce the dot pattern by adjusting the cutoff frequency, and to remove the dot pattern appearing in the color difference signal, adjust the weight according to the transition region detection signal of luminance and the dot pattern detection signal of the color difference signal, and The dot pattern is reduced by adjusting the cutoff frequencies of the vertical low pass filters 634 and 635.

7 illustrates the operation of the transition region detectors 622 and 632 as follows.

Here, the transition region detection method for the luminance and the chrominance signal is the same.

As the component signals Y or I / Q are input, the signals a to i shown in FIG. 5 are stored by the two line memories LM1 and LM2 and the plurality of sample memories R. As shown in FIG.

Here, the pixel for the current interference cancellation process is 'e'.

In the case of the horizontal transition region detection on the pixels stored as described above, the horizontal transition region determination signal (luminance: , Color difference: , )

--------- (5)

Here, the horizontal transition region determination signal ( ) For luminance , For color difference signal Wow Means.

The meaning of the calculation as in Equation (5) is that the brightness difference between the left and right pixels d and f of the pixel e currently being subjected to interference cancellation is a threshold value ( In other words, when it is determined that there is a horizontal transition and there is a horizontal transition in at least one of the two scanning lines above and below, the pixel e is determined to exist in the horizontal transition region.

Therefore, the horizontal transition area determination signal of the color difference signal ( , ) When one or more of the signals is '1', the horizontal transition area determination signal of the color difference signal ( ) To '1'.

That is, the transition region detectors 622 and 632 The horizontal operation is detected by performing the same operation.

In the case of detecting the vertical transition region, a vertical transition region determination signal (luminance: , Color difference: , )

--------- (6)

Here, the vertical transition region determination signal ( ) For luminance , For color difference signal Wow Means.

The meaning of the calculation as in Equation (6) is that the brightness difference between the upper and lower pixels b and h of the pixel e currently being subjected to interference cancellation is a threshold value ( If it is determined that there is a vertical transition larger than), and that at least one vertical transition exists to the left and right of the pixel e, the pixel e is determined to exist in the vertical transition region.

Therefore, the vertical transition region determination signal of the color difference signal ( , ) When one or more of the signals is '1', the horizontal transition area determination signal of the color difference signal ( ) To '1'.

That is, the transition region detectors 622 and 632 A vertical transition region is detected by performing the same operation.

FIG. 8 illustrates the operation of the dot pattern detectors 621 and 631.

Here, the dot pattern detection method for the luminance and the color difference signal is the same.

As the component signals Y or I / Q are input, the pixels a to i shown in FIG. 5 are similar to the transition region detectors 622 and 632 by the two line memories LM1 and LM2 and the sample memories R. As shown in FIG. Save it.

Here, the pixel currently subjected to interference cancellation is 'e'.

In the case of horizontal dot pattern detection on the pixels stored as described above, the horizontal dot pattern detection signal (luminance: , Color difference: , )

-------- (7)

Here, the horizontal dot pattern determination signal ( ) For luminance , For color difference signals , Means the threshold ( ) Is the threshold ( About 2 times is appropriate.

The meaning of the calculation as shown in Equation (7) is that the pixel (e), which is currently subjected to interference cancellation, has a horizontal dot pattern form as shown in FIG. 5, and the brightness difference between pixels (a) and (c) in the upper scanning line is small or lower. When the difference in brightness of the pixels g and i on the scan line is small, it means that the pixels e are included in the horizontal dot pattern.

In the case of vertical dot pattern detection, a vertical dot pattern detection signal (luminance: , Color difference: , )

----------------- (8)

Here, the vertical dot pattern determination signal ( ) For luminance , For color difference signal , Means.

The meaning of the operation as shown in Equation (8) is that the pixel (e) that is currently subjected to the interference cancellation process has a vertical dot pattern form as shown in FIG. When the difference in brightness is small, it means that the pixel e is included in the vertical dot pattern.

9 conceptually illustrates the operation of the horizontal and vertical low pass filters 624 and 634 and 625 and 635.

Here, the processing of the luminance Y and the color difference signals I, Q is performed in the same process.

In FIG. 5, when the position of the pixel currently being processed to remove noise is the pixel e, vertical filtering is performed as shown in Equation 9 below.

------------------- (9)

here, Is a line comb filter and performs the same operation as Equation (10) below.

--------------------- (10)

The pixels obtained by the low pass filtering in the vertical direction as in Equation (9) are finally removed by the low pass filtering in the horizontal direction as in Equation (11) below.

----------------- (11)

Applied to equation (9) (11) above , Is a weight determined by the vertical and horizontal dot pattern detection results and the transition region detection results, respectively, and has a value of 0.1 to 1.0.

That is, the weight has a value close to '1' when the dot pattern and the transition region are detected, and conversely, has a value close to '0' when it is not detected.

The order of applying the horizontal filtering process and the vertical filtering process is irrelevant to the calculation result.

That is, the result of applying the horizontal filtering after the vertical filtering process and the result of applying the vertical filtering after the horizontal filtering process have the same value.

However, since the line memory is required for the filtering in the vertical direction, it is advantageous to implement the filtering in the horizontal direction after the filtering in the vertical direction.

The above process is the same as the process of the color difference signal I / Q as the process of the luminance Y.

Accordingly, the vertical / horizontal (V / H) low pass filters 624 and 634 and 625 and 635 are configured as shown in FIG. 10 and the operation thereof will be briefly described as follows.

First, an operation on the luminance signal Y will be described.

The vertical low pass filter 624 outputs the low pass filtered pixel value in the vertical direction to the horizontal low pass filter 625 by the same operation as the equation (9).

The horizontal low pass filter 625 delays the pixel value output from the vertical low pass filter 624 by the delay element R so that each pixel value ( ) ( ) ( )

Thereafter, the horizontal low pass filter 625 outputs the luminance signal y 'by performing the horizontal low pass filtering process by the operation as in Equation (11).

The configuration of the vertical / horizontal low pass filters 625 and 635 for the chrominance signal is the same as that of the vertical and horizontal low pass filters 624 and 634 for the luminance signal.

Meanwhile, weights inputted from the weight generators 623 and 633 to the vertical / horizontal low pass filters 624 and 634 and 625 and 635. ) ( ) Are the same calculation method, and the weight calculation method for the luminance signal (Y) and the color difference signal (I, Q) is also the same.

For example, the horizontal dot noise of the luminance signal is generated when there is a vertical transition of the color signal and a horizontal dot pattern of the luminance signal, so that the weight applied to the horizontal low pass filter 625 for removing the dot pattern of the luminance signal ( ), The horizontal dot pattern detection result of the luminance signal obtained by the above equation (7) ) And the detection result of the vertical transition region of the color signal obtained by Equation (5) ) And the weight ( ).

Similarly, the weight applied to the vertical low pass filter 624 of the luminance signal ( ), The vertical dot pattern detection result of the luminance signal ( ) And the horizontal transition region detection result of the color signal ( ) And the weight ( Generates.

The weight generators 623 and 633 calculate the weight W as shown in the truth table of FIG. 11 (b) by the weight generation logic 641 as shown in FIG.

That is, the weight applied to the horizontal low pass filter 625 for the luminance signal ( ) Is '0' Since '0.0' is the horizontal low pass filter 625 performing the operation as shown in Figure 9 output ( )silver This means that the horizontal dot noise is not removed.

if, Is '1' Is 1, it is highly likely that the luminance signal contains horizontal dot noise. In case of the horizontal low pass filter 625 which becomes '1.0' and performs the operation as shown in FIG. )silver To eliminate horizontal dot noise.

However, in the weight calculation method of FIG. 11, since the weights applied to the vertical / horizontal low pass filters 624, 634 and 625, 635 may be significantly different for each pixel, the consistency of noise reduction may be reduced, which may result in deterioration of image quality.

Therefore, the weight generators 623 and 633 are configured as shown in the block diagram of FIG. 12 to introduce the concept of IIR filtering to improve the dot noise elimination consistency.

The weight generators 623 and 633 configured as shown in FIG. 12 have weights outputted from the weight generation logic 641 of FIG. ) And the output of the delay element 645 ( ) Is a constant ( Weighted to generate a weight W to be applied to the current pixel.

That is, the weight of the weight generation logic 641 in the multiplier 642 ( ) To a constant ( ) And the output of the delay element 645 in the multiplier 643 ( ) To a constant ( ), The adder 644 adds the output values of the multipliers 642 and 643 to generate a weight W applied to the current pixel.

The weight W obtained through the above process is fed back to the delay element 645 and output to the vertical / horizontal low pass filters 624, 634 and 625, 635.

Where constant ( ) Has values from '0' to '1', so constants ( ) Is closer to '1', the weight obtained by the dot determination signal D and the transition region determination signal T ) Will be affected.

On the other hand, constants ( ) Is closer to '0' and is affected by the weight (W) applied to the previous pixels. In this case, the coherence of the noise reduction is increased, but the accuracy of the determination of the presence or absence of dot noise for the current pixel is inferior.

Therefore, the constant ( ) Is appropriate to apply about 0.5.

The delay element 645 is a register when calculating the weight applied to the horizontal low pass filters 625 and 635 and a line memory when calculating the weight applied to the vertical low pass filters 624 and 634.

Accordingly, when the weight generators 623 and 633 are configured as shown in FIG. 12, a line memory for storing the output weight W represented by a large number of bits is additionally required. Therefore, when implemented by an ASIC or the like, the number of gates is greatly increased.

In order to improve the gate count increase problem, the weight generators 623 and 633 are configured as shown in the block diagram of FIG.

The weight generators 623 and 633 as shown in FIG. 13 have a 2-bit delay element 653 to calculate a weight.

The dot determination signal D and the transition region determination signal T are inputted to the weight generation logic 641 which performs the operation as shown in the truth table of FIG. ) And its output ( ) And the output of the 2-bit delay element 653 ( ) Are input to the weight read logic 651 which performs the operation as shown in the truth table of FIG. 14 (b) and the weight write logic 652 which performs the same as the truth table in FIG. 14 (c).

Accordingly, the weight read logic 651 outputs the weight generation logic 641 and the output of the 2-bit delay element 653. ) ( A weight W having a value of '0.0' to '1.0' is output as a weight of the vertical and horizontal low pass filters 624, 634 and 625 and 635, as shown in FIG.

In addition, the weight write logic 652 is inputted from the weight generation logic 641 and the 2-bit delay element 653. ) ( Value to be used to find the weight W at the next pixel according to the value of ) Is stored in the 2-bit delay element 653.

After all, if the weight (W) applied to the previous pixel is close to '1', the output ( ) Becomes a value close to '3', so that the weight W shown in the truth table of FIG. 14 (b) becomes a large value and is determined by the dot determination signal D and the transition region determination signal t. Output for the current pixel ( ) Becomes a large value by the same operation as the truth table in Fig. 14 (a), the weight W becomes a large value as shown in the truth table in Fig. 14 (b).

Therefore, the output ( ) ( ) Are '3' and '2' respectively, the weight W for the current pixel is '1' and the output ( ) ( When both are '0', the weight W becomes '0'.

The output of the weight write logic 652 is an input ( ) ( 4 types of values of 0, 1, 2, and 3 are output according to the value of?). Therefore, the delay element 653 only needs to store 2-bit information.

However, since the dot noise appearing in the decoded NTSC signal is generally included in the luminance, the weight generators 623 and 633 can be implemented as shown in FIGS. 11, 12, and 13, but considering the complexity and performance of the implementation. It is preferable to use the same method as in Fig. 13 for the luminance signal and the same method as in Fig. 11 for the color difference signal.

As described in detail above, the present invention can remove the dot pattern appearing at the edge of the demodulated NTSC color TV signal due to incomplete luminance / color signal separation in the NTSC decoder, thereby reducing image quality degradation.

The present invention can be applied to most image-related applications using the component image obtained by the conventional NTSC TV camera to eliminate the dot pattern caused by the luminance / color signal interference.

In particular, the present invention has an effect of reducing the deterioration of image quality due to the inaccurate dot noise removal process by improving the consistency of dot noise removal by introducing the IIR concept for weight generation.

In addition, when the image format is 4: 2: 2, the configuration is simplified because only four line memories are required to remove all the dot patterns appearing in the luminance and color difference signals.

Claims (12)

NTSC decoder which removes dot pattern included in NTSC composite video signal (CV) and demodulates component video signal (YIQ) by NTSC decoding, and removes interference noise included in component video demodulated by this NTSC decoder. Dot pattern removal filter, characterized in that it comprises a noise canceller. The dot pattern of claim 1, wherein the noise canceller is configured to adjust a cutoff frequency for vertically and horizontally low pass filtering by adjusting weights according to transition region detection of the color difference signal and dot pattern detection of luminance. Dot pattern removal filter, characterized in that to reduce the. The low pass filtering of claim 1 or 2, wherein the noise canceller passes a line memory temporarily storing a luminance / color difference signal, and a low pass filtering unit passing only predetermined low-pass components in the vertical and horizontal directions with respect to the luminance signal delayed by two lines in the line memory. Means, a dot pattern detector for detecting a dot pattern in the horizontal and vertical directions with respect to the luminance signal for each pixel by receiving the luminance signal from the line memory, and a horizontal and vertical transition for receiving the color difference signal from the line memory. A weighted signal for controlling a cutoff frequency of the vertical and horizontal low pass filters by inputting a transition region detector for detecting an area, a dot pattern in the dot pattern detector, and transition region information in the transition region detector; A dot pattern removal filter comprising a vertical / horizontal weight generator. 2. The dot of claim 1, wherein the noise canceller is configured to adjust a cutoff frequency for vertically and horizontally low pass filtering by adjusting weights according to a transition region detection of a luminance signal and a dot pattern detection of a color difference signal. Dot pattern removal filter, characterized in that to reduce the pattern. 5. The low pass filtering of claim 1 or 4, wherein the noise canceller passes a line memory temporarily storing a luminance / color difference signal and a low pass filter passing only a predetermined low-pass component in the vertical and horizontal directions with respect to the color difference signal delayed by two lines in the line memory. Means, a dot pattern detector for detecting a dot pattern in the horizontal and vertical directions with respect to the color difference signal for each pixel by receiving the color difference signal from the line memory, and a horizontal and vertical transition for receiving the luminance signal from the line memory. A transition region detector for detecting a region, a weight signal for controlling the cutoff frequency of the vertical and horizontal low pass filters by inputting a dot pattern in the dot pattern detector and transition region information in the transition region detector; A dot pattern removal filter comprising a vertical / horizontal weight generator. The noise canceller of claim 1, wherein the noise canceller adjusts a cutoff frequency for vertical / horizontal low pass filtering by adjusting weights according to the transition region detection of the color difference signal and the dot pattern detection of luminance, and the transition region detection of the luminance and the color difference signal. A dot pattern removal filter, characterized in that to reduce the dot pattern appearing in the luminance signal and the chrominance signal by adjusting the weight according to the dot pattern detection of the to adjust the cutoff frequency for low pass filtering in the vertical / horizontal direction. 7. The first low pass filtering of claim 1 or 6, further comprising: a line memory for delaying the luminance / color difference signal by two lines and only a predetermined low pass component in the vertical and horizontal directions with respect to the luminance signal delayed by two lines in the line memory. Means, a first dot pattern detector for detecting a dot pattern in the horizontal and vertical directions with respect to the luminance signal by receiving the luminance signal from the line memory, and a horizontal and vertical direction receiving the color difference signal from the line memory; A cutoff frequency of the first low pass filtering means is input by inputting a first transition region detector for detecting a transition region of the first transition region, a dot pattern of the first dot pattern detector, and transition region information of the first transition region detector. A first weight generator for outputting a weighted signal for control and a color difference signal delayed by two lines in the line memory; Second low-pass filtering means for passing only a predetermined low-pass component in the vertical and horizontal directions, and a second dot pattern for detecting a dot pattern in the horizontal and vertical directions with respect to the color difference signal for each pixel by receiving a color difference signal from the line memory; A second transition region detector for detecting a transition region in a horizontal and vertical direction by receiving a color difference / luminance signal from the line memory, a dot pattern in the second dot pattern detector, and a second transition region detector And a second weight generator for outputting a weighted signal for controlling a cutoff frequency of the second low pass filtering means by inputting transition region information of the second low pass filtering means. A first step of removing the dot pattern included in the composite video signal and separating and outputting the component signal as a component signal by NTSC decoding; and a second step of detecting the presence or absence of a transition region with respect to the output component signal; A third step of detecting the presence or absence of a dot pattern with respect to the signal, and a fourth step of performing low-pass filtering in the vertical and horizontal directions with respect to the component signal by referring to the detected transition region and the dot pattern. How to remove a pattern. The component according to claim 8, wherein the second step is assuming that the pixel (a, b, c) (d, e, f) (g, h, i) is in the line and the pixel to be subjected to interference cancellation is 'e'. The brightness difference between the left and right pixels d and f is based on a threshold value based on the pixel e that is currently subjected to interference cancellation with respect to the signal. A first process of determining whether there is a horizontal transition in at least one of the two scan lines larger than and larger than), and if it is determined that the pixel e is in the horizontal transition region, The brightness difference between the upper and lower pixels b and h is determined based on the second process of activating and the pixel e currently being subjected to interference cancellation. ) And a third process of determining whether there is at least one vertical transition on the left and right, and a fourth process of activating the vertical transition region determination signal when it is determined that the pixel e is in the vertical transition region in the third process. Dot pattern removal method characterized in that. The component according to claim 8, wherein the third step is assuming that the pixel (a, b, c) (d, e, f) (g, h, i) is in the line and the pixel to be subjected to interference cancellation is 'e'. The pixel (e) that is currently subjected to interference cancellation with respect to the signal has a horizontal dot pattern shape and the brightness difference of the pixels (a) (c) on the upper scanning line is small or the brightness difference of the pixels (g) (i) on the lower scanning line Is smaller than the first step of determining whether the pixel e is included in the horizontal dot pattern, and if it is determined that the pixel e is included in the horizontal dot pattern in the first step, the horizontal dot pattern determination signal ( A second process of activating a pixel and a pixel c in a lower scan line or a lower brightness difference between pixels (a) and (g) among left and right pixels having a vertical dot pattern form a third step of determining whether the pixel e is included in the vertical dot pattern because the brightness difference of (i) is small; and if the pixel e is included in the vertical dot pattern, the vertical dot pattern determination signal ( ) Is a fourth process of activating the dot pattern removal method. The method of claim 8, wherein the fourth step includes calculating a weight value according to the detected transition region and the dot pattern with respect to the luminance and color difference signals, and blocking for vertical and horizontal low pass filtering based on the calculated weight value. A dot pattern removing method comprising the step of adjusting the frequency. 12. The method of claim 11, wherein the weights calculate an IIR filtering scheme.