KR19980017212A - Improved Image Decoding System - Google Patents

Abstract

The present invention calculates the complexity of the image restored to the original signal based on the occasional quantization parameter information that determines the predetermined quantization step size in each macroblock unit, and according to the calculation result, the high frequency of the image signal restored to the original signal An improved image decoding system suitable for suppressing image quality deterioration due to quantization error by adaptively removing a component, for which the present invention reconstructs by averaging a plurality of quantization parameter values for a currently reconstructed frame. An average quantization parameter calculation block for calculating an average quantization parameter value of one frame; The complexity of the currently reconstructed frame is calculated by comparing the average quantization parameter value of the calculated one frame with a preset reference value, and based on the calculated complexity information, the complexity is present immediately after the frame used for the complexity calculation. A control block for generating a filtering control signal for the recovered next frame; Band limiting means for generating a reconstructed frame from which high frequency components are removed by filtering a next frame reconstructed based on a filter coefficient determined according to the generated filtering control signal to limit the pass band; And at least two paths for providing the reconstructed frame signal to the digital / analog conversion means, and generated based on a result of comparing the average quantization parameter value of the calculated one frame provided from the control block with a predetermined reference value. And a switching block for providing the reconstructed frame signal to the digital / analog converting means in response to the switching control signal to the digital / analog converting means, or for providing the reconstructed frame signal from which the high frequency component is removed to the digital / analog converting means. Image quality deterioration due to errors can be effectively reduced.

Description

Improved Image Decoding System

The present invention relates to a video decoding system for decoding an encoded video signal. More particularly, when decoding a compressed coded video bitstream, received quantization parameter information for determining a predetermined quantization step size in units of macroblocks. The present invention relates to an improved image decoding system suitable for compensating for image quality degradation caused by quantization error after decoding with reference to the complexity of an image signal reconstructed into an original signal.

As a general method for decoding a compression-encoded video signal, MPEG-1 and MPEG-2 (the official name of this Recommendation is ITU-T Rec. H222.0 | ISO / IEC 13818) are widely known. Such a recommendation is to design an apparatus for decoding according to this format since the input stream conforms to the format prescribed by MPEG at the time of decoding the video signal.

However, in this process, since a decoding method is generally recommended, an implementation method and an apparatus of an efficient decoder at the time of decoding are still actively researched everywhere.

1 is a block diagram of such a typical image decoding system. As shown in the figure, a bit stream of an encoded video signal that is separated and input through a system decoding system (not shown) is input to and stored in the reception buffer 101, and thus is stored in the reception buffer 101. Is provided to the variable length decoding unit 102 at predetermined time intervals in synchronization with the decoding timing to undergo variable length decoding, which is an inverse process of variable length coding in the transmission-side video encoding system.

The quantized stream (ie, quantized DCT transform coefficient) and motion vectors (motion displacement) obtained here are separated, and the separated quantized stream (quantized DCT transform coefficient) is separated from the inverse quantizer 103 of the next stage. ) And inverse quantized, and the motion vectors are provided to the motion compensation unit 104 to be described later for motion compensation.

Then, the DCT transform coefficients dequantized by the inverse quantization unit 103 are provided to the inverse DCT unit 105. In this inverse DCT unit 105, the DCT definition inverse process in the transmitting image coding system, that is, The inverse DCT transform is performed using an inverse discrete cosine transform to convert an original signal before compression coding (that is, a motion signal using the current frame and the previous frame and a differential signal using compensation). That is, the signal reconstructed by the original signal output from the inverse DCT unit 105 is a difference signal between the prediction frame and the current frame obtained through motion estimation and compensation between the current frame and the previous frame.

Accordingly, in the image decoding system, the difference signal from the inverse DCT unit 105 and the motion compensated frame signal provided from the motion compensation unit 104 described later (that is, the motion vectors are used to the frame memory 107). From the reconstructed previous frame provided from the () reconstructed by the adder 106 to create an image frame restored to the original signal.

Next, the image frame signal from the adder 106 restored to the original signal is stored in the frame memory 107, which is converted into an analog signal through the D / A converter 108 for display. And then to the display side, not shown. In addition, the reconstructed image frame signal stored in the frame memory 107 is provided to the motion compensator 104 for motion compensation of the next frame that is temporally continuous. Therefore, since the image frame reconstructed through this process is continuously provided to the motion compensator 104, motion compensation between frames that are continuously temporally performed can be performed.

Meanwhile, the motion compensator 103 compensates for the movement of the current frame from the previous frame based on the motion vector (or displacement) to the previous frame and the variable length decoder 102 stored in the frame memory 107, that is, the frame memory. A reconstructed frame signal obtained by moving the reconstructed previous frame provided by 107 by a motion vector is generated, and the reconstructed frame signal is provided to the adder 106 as a motion compensated frame signal.

Therefore, the adder 106 adds the motion compensated image frame signal from the motion compensator 104 and the difference signal from the inverse DCT unit 105, thereby generating a continuous reconstructed frame signal to generate a frame. To the memory 107.

On the other hand, a general video signal decoding system of the type described above is well mentioned in the existing MPEG-1 and MPEG-2 recommendations.

The encoded bit stream received through the reception channel and stored in the reception buffer 101 starts decoding when the decoding time comes. The decoding time is determined by extracting a specific parameter transmitted from the image encoding system of the transmitting side. The time point is related to the size of the reception buffer 101 and is determined so that a malfunction (overflow, underflow) of the reception buffer 101 does not occur.

For example, assuming that there is no error in the received video bit stream and that the stream is an ideal channel that is input at the same rate in time, no malfunction of the receive buffer 101 will occur.

However, the amount of bits generated during encoding in the transmission-side video encoding system is inevitably different for each frame. In other words, in the case of a video having a large complexity, a relatively large amount of bits is generated after encoding, and in the case of a simple image, a relatively small amount of bits is generated after encoding. By adjusting the quantization step size in consideration of the transmission rate and the like, the bit generation is appropriately adjusted.

In other words, if the video inputted for encoding in the encoding system on the transmitting side is a relatively complex video, the amount of bits generated after encoding is increased. In the video encoding system, the quantization step size is adjusted in consideration of the capacity and transmission rate of the transmission buffer. The bit generation amount is limited by increasing the quantization parameter (QP) value. On the contrary, when the image input for encoding in the encoding system on the transmitting side is a relatively simple image, the bit generation amount after encoding is reduced. In this case, the bit generation amount is controlled by reducing the quantization parameter (QP) value. do.

On the other hand, as described above, when the video signal quantized according to the quantization parameter determined in consideration of the bit generation amount and the transmission rate due to the complexity of the image is quantized with a large value quantization parameter, Image quality deterioration, that is, image quality deterioration due to a noticeable blocking phenomenon is caused by dividing the screen finely. In particular, the quantization error is a lot of high-frequency components that are relatively insensitive to the human visual characteristics, if the image is reproduced as it will inevitably be worse image quality deterioration. Therefore, if the quantization error mainly distributed in such high frequency components can be eliminated, it may be possible to reproduce an image having more natural image quality.

Accordingly, the present invention devised in view of the above point, and calculates the complexity of the image reconstructed with the original signal based on the received quantization parameter information for determining the predetermined quantization step size in each macro block unit, and calculates the calculated It is an object of the present invention to provide an improved image decoding system capable of suppressing image quality degradation due to quantization error by adaptively removing high frequency components of an image signal after being restored to an original signal according to the result.

In order to achieve the above object, the present invention, by applying a variable length decoding, inverse quantization, and inverse DCT and motion compensation technique to the encoded video bit stream to restore the original signal before encoding and the restored frame signal analog A digital / analog conversion means for converting a signal into a display side and providing the signal to a display side, wherein the quantization is performed using a quantization parameter of each macroblock unit received together with the video bit stream. An average quantization parameter calculation block for averaging a plurality of quantization parameter values in macroblock units for one frame to calculate an average quantization parameter value of one frame; The complexity of the currently reconstructed frame is calculated by comparing the average quantization parameter value of the calculated one frame with a preset reference value, and immediately based on the calculated complexity information, in time rather than the frame used for calculating the complexity. A control block for generating a filtering control signal for the restored next frame which exists later; Band limiting means for generating a reconstructed frame from which high frequency components are removed by filtering the reconstructed frame based on the filter coefficient determined according to the generated filtering control signal and limiting the pass band; And at least two paths for providing the reconstructed frame signal to the analog conversion means, based on a result of comparing the average quantization parameter value for the calculated one frame provided from the control block with a preset reference value. And a switching block for providing the reconstructed frame signal to the digital / analog converting means or the reconstructed frame signal from which the high frequency component is removed to the digital / analog converting means in response to a switching control signal generated. It provides a video decoding system.

1 is a block diagram of a conventional video decoding system.

2 is a block diagram of an improved image decoding system according to a preferred embodiment of the present invention.

3 is a view illustrating a determination region for high frequency component limitation determined as one fixed level for an 8x8 pixel block as an example when a reconstructed image is a complex image according to the present invention;

4 is a view showing a decision region for limiting high frequency components that is adaptively determined based on the complexity of an 8x8 pixel block as an example when a reconstructed image is a high complexity image.

5 is an illustration of an 8x8 pixel block as an example in accordance with the present invention;

6 is an exemplary view showing a one-dimensional low-pass filter coefficient having a dimension of 7 as an example according to the present invention.

7 is an exemplary view illustrating a horizontal filtering process at (0, 4) and a vertical filtering process at (3, 0) as an example according to another embodiment of the present invention.

FIG. 8 is an exemplary diagram illustrating a two-dimensional low pass filter coefficient of 3x3 order as an example according to another embodiment of the present invention. FIG.

9 is an exemplary diagram illustrating a two-dimensional filtering process at position (3,3) as an example according to another embodiment of the present invention.

FIG. 10 is a detailed block diagram of a band limiting block shown in FIG. 2 according to another embodiment of the present invention. FIG.

Explanation of symbols on the main parts of the drawings

101: Receive buffer 102: Variable length decoder

103: reverse quantization unit 104: motion compensation unit

105: Reverse DCT106: Adder

107: frame memory 108: D / A converter

210: AQP output block 220: Control block

230: band limit block 240: switching block

1141: DCT block 1143: quantization block

1145: frequency selector 1147: dequantization block

1149: IDCT block

The above and other objects and various advantages of the present invention will become more apparent from the preferred embodiments of the present invention described below with reference to the accompanying drawings by those skilled in the art.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

According to the present invention, if it is determined that the video signal reconstructed as the original signal before encoding through variable length decoding, inverse quantization, inverse DCT, etc. has a complex image or a simple projection, it is determined that the image signal is a complex image. In order to prevent image quality degradation due to quantization error by adaptively removing high frequency components relatively insensitive to human visual characteristics, the complexity of the reconstructed image signal is calculated using macroblock quantization parameters.

That is, if the average quantization parameter for each frame is larger than the predetermined reference value, it is determined as a complex image and adaptive filtering is applied to the reconstructed image signal. Signal) is transmitted to the display side, and if it is smaller than the reference value, it is determined as a simple image and is directly transmitted to the display side without applying adaptive filtering to the reconstructed video signal.

This is because the encoding system on the transmitting side encodes by increasing the quantization parameter (the quantization parameter QP has an integer value of 1 to 31) for the image that is complicated at the time of quantization of the image and the bit generation amount after encoding is determined to be large. It is based on lowering the amount of subsequent bit generation. Therefore, it will be very reliable to calculate the complexity of the reconstructed image based on the quantization parameter according to the present invention.

2 shows a block diagram of an improved image decoding system according to a preferred embodiment of the present invention. The image decoding system of the present invention shown in FIG. 1 includes a reception buffer 101, a variable length decoder 102, an inverse quantization 103, a motion compensator 104, an inverse DCT 105, an adder 106, and The frame memory 107, the D / A converter 108, the AQP calculation block 210, the control block 220, the band limiting block 230, and the switching block 240 are included.

As can be seen from Figure 2, the video decoding system of the present invention, the conventional video decoding system shown in Figure 1 AQP characteristic calculation block 210, control block 220, band limit block 230 and switching In addition to the configuration of the block 240, the main configuration feature, and by the additional components to achieve the object of the present invention, that is, to suppress the image quality degradation caused by the quantization error when the reconstructed image is a complex image The goal will be achieved.

Accordingly, as described above, the remaining members except for the members 210, 220, 230, and 240 added to the conventional decoding system exhibit substantially the same components that perform the same functions as those of the conventional decoding system. Detailed description herein is omitted in order to avoid description.

Referring to FIG. 2, in the average quantization parameter (AQP) calculation block 210, the macro block unit provided by the inverse quantization unit 103 described above is input through the line L11 to input the quantization parameters (QPs) to average each frame. A quantization parameter (AQP) value is calculated, and the calculated AQP value is provided to the next control block 220. In this case, if the current reconstructed video signal is a relatively complex image, the AQP value will have a large value, and in the case of a simple image, the AQP value will have a large value. In addition, the video signal does not change rapidly every frame due to its characteristics. In the present invention, the AQP value calculated in the currently reconstructed frame is used to calculate the image complexity of the next frame using this characteristic. That is, in the present invention, the AQP value of the nth frame is used as the basis for calculating the complexity of the n + 1th frame.

That is, assuming that one frame video signal has a horizontal size and a vertical size of M * N, and one macroblock has a size of L * L (for example, 16x16), one frame data is restored. After that, the number of generated QP becomes (M / L) * (N / L). As an example, assuming 16 × 16 macroblocks for an input image of 352 * 288, the number of macroblocks for a frame is (352/16) * (288/16), which is 16 × 18, that is, 288. will be. Accordingly, the calculated AQP value for one frame is used as the complexity of the image signal to be reconstructed next. In this case, a large AQP value means a complex image, and a small AQP value means a simple image.

Meanwhile, in the control block 220, is the next frame to be restored based on the AQP value provided from the AQP calculation block 210, ie, the AQP value obtained by averaging the QP values of one frame, is a frame having a complex shape? Or determining whether the frame has a simple image, and generating a filtering control signal B on the line L13 and a switching control signal, which is a logic signal of a high or low level, on the line L15 based on the determination result. .

That is, if the AQP value from the AQP calculation block 210 is greater than the preset reference value, the control block 220 determines that the next frame of the frame used for calculating the APQ value has a complex image, and the AQP value is preset. If it is smaller than the reference value, it is determined that the frame immediately following the frame used for calculating the AQP value has a simple image. Here, for example, since the QP value has an integer value between 1 and 31, it may be desirable to set the reference value to about 16, which is the intermediate value thereof.

Therefore, if it is determined that the input AQP value is larger than the reference value and is a frame having a complex image, the control block 220 generates a filtering control signal B on the line L13 to control the filtering flood in the band limiting block 230. At the same time, a switching control signal (a logic signal having a high or low level) is generated on the line L15 to control the switching operation in the switching block 240, that is, the contact is connected to ab while the contact is connected to ac. .

Meanwhile, in the band limiting block 230, the peeling on the line L13 performs an adaptive band limiting process, that is, filtering on the restored frame signal provided from the frame memory 107 based on the control signal B, thereby setting a predetermined high frequency component. This removed reconstructed frame signal is generated on the line L17. That is, when filtering the reconstructed frame signal, in the band limiting block 230, as an example, as shown in FIG. 3, Z (1,7) and Z (2,6) or less are all zeros (0). Mapping is used to limit the bandwidth and filter.

Then, the reconstructed frame signal from which the high frequency component is removed is output to the D / A converter 108 of the next stage via the contact b-a of the switching block 240. Accordingly, the D / A converter 108 may convert the reconstructed frame signal from which the high frequency component has been removed into an analog signal and then provide it to the display side, not shown, for display on a monitor.

On the other hand, in the control block 220, when the band limiting block 230 filters the restored frame signal provided from the frame memory 107 in accordance with the present invention, a fixed one is generated. The filtering control signal B can be calculated by the following equation so as to adaptively (or selectively) filter the reconstructed frame without filtering the reconstructed frame by the level (that is, the filter coefficient) of.

B = 1 if (eMIp (0) = Me + 0.5 * Se

= 2 else if (eMIp (0) = Me + 1 * Se

= 3 else if (eMIp (0) = Me + 1.5 * Se

= 4 else

As a result, the band limiting block 230 performs appropriate band limiting processing, i.e., adaptive filtering, on the recovered frame signal provided from the frame memory 107 based on the filtering control signal B on the line L13. Perform. That is, when filtering the reconstructed frame signal, the band limiting block 230 corresponds to the filtering control signal B from the control block 220 based on the complexity of the input image, as shown in FIG. 4 as an example. By mapping the value below the output value (that is, the value below the dotted line in FIG. 4) to a zero value, the bandwidth is limited, that is, filtering is performed. In other words, according to the present embodiment, the high frequency component removal level of the recovered frame signal is selectively (or adaptively) adjusted according to the complexity of the recovered frame.

Then, as described above, the reconstructed frame signal from which the high frequency component is adaptively (or selectively) removed is output to the D / A converter 108 through the b-a path of the switching block 240.

In addition, in the control block 220, if it is determined that the input AQP value is larger than the reference value and is a frame having a complex image, a high level switching control signal is generated on the line L15 so that the contact point of the switching block 240 is ca to ba. Control to switch to Therefore, the reconstructed frame signal from which the high frequency component is removed through the band limit block 230 is output to the D / A converter 108 of the next stage.

On the other hand, when the band limiting block 230 removes the high frequency components from the frame signal reconstructed through the filtering according to the present invention, a method for removing the high frequency components includes a one-dimensional low pass filtering technique and a two-dimensional low pass filtering. A band limiting method using a Discrete Fourier Transform (hereinafter abbreviated as DFT) and a band limiting method using a Discrete Cosine Transform (abbreviated as DCT).

As described above, the one-dimensional low pass filtering technique, which is one of techniques for removing high frequency components insensitive to human visual characteristics, is performed by multiplying a low pass filter by an input reconstructed frame (image) signal. do. That is, when f (x, y) is the value of each pixel for an image of one block of N × M (for example, 8 × 8 block), f (x, y) is one block as shown in FIG. 5 as an example. In the case of a block of x8, the position values x and y of the pixels in the horizontal and vertical directions have integer values between 0 and 7, and each value has a level value between 0 and 255. That is, as can be seen from Fig. 5, the position values in the horizontal and vertical directions of each pixel of the 8x8 block have a value of f (0, 0) to f (7, 7).

On the other hand, as the one-dimensional low pass filter in the present invention, as shown in FIG. 6 as an example, it is assumed that the one-dimensional low pass filter coefficient has seven orders. The low pass filter is an example of a low pass filter that band-limits the signal to have a frequency bandwidth of fs / 4 since the frequency bandwidth is fs / 2 when the sampling frequency of the input video signal is fs.

Therefore, in the band limiting block 230, the process of performing the one-dimensional low pass filtering on the reconstructed frame signal according to the present invention is a one-dimensional low pass for each direction since the reconstructed frame signal is a two-dimensional signal in the horizontal and vertical directions. This can be achieved by performing filtering. This process is illustrated in detail in FIG. 7, which illustrates a horizontal filtering process at (0,4) and a vertical filtering process at (3, 0). That is, assuming that the signal z (x, y) low-pass filtered in the vertical direction at the position of (x, y), this is calculated by the following equation.

[Equation 2]

In Equation 2, since T means the order of the filter, T = 7. Thus, the u value has an integer value between -3 and 3. Also, in Equation 2, k (u) is a filter coefficient value, and f (x, y) is a pixel value. If the (yu) value of f (x, yu) in Equation 4 is less than 0, the value is 0. If the value is larger than the M-1 value corresponding to the image size of one frame, the value is M-1. Do it. This is a method of setting the end value when the pixel value exists only in the area, that is, 0 to M-1, and this filtering method is a technique known in the art.

Therefore, if the filtering is performed at all pixel positions as in Equation 2, as shown in FIG. 7, one-dimensional filtering in the horizontal and vertical directions may be obtained. In this case, in the process of performing such filtering at all pixel positions, horizontal filtering may be performed first, and vertical filtering may be performed again or the order may be performed on the horizontal filtered result.

On the other hand, in the one-dimensional low-pass filtering as described above may be configured to completely remove or partially remove the high-frequency components of the recovered frame signal using only a predetermined (or fixed) filter coefficient, In consideration of the degree of complexity, a plurality of filter coefficients (for example, four, etc.) may be used to adaptively (or selectively) remove high frequency components of the reconstructed image according to the magnitude of the complexity. In this case, when adaptively filtering a reconstructed frame using a plurality of filter coefficients, hardware implementation may be more complicated than image filtering using a single predetermined filter coefficient, but precise control of image reproduction is performed. Has another advantage that it is possible. Therefore, in the case of adaptive filtering using a plurality of filter coefficients as described above, it may be selectively applied according to its application range.

On the other hand, two-dimensional low pass filtering among the techniques for the high frequency component removal of the reconstructed frame, as shown in Fig. 5, is performed on an image of one block of N × M (eg, 8 × 8 blocks). Assuming that each value of each cell is f (x, y), and as an example, as shown in Fig. 8, the two-dimensional low pass filter coefficient is assumed to have a 9th order having three orders, As determined by the filter coefficient k, the value of k (0, 0) is 1/2, and other values have a value of 1/16.

Accordingly, the process of performing 2D low pass filtering on the reconstructed frame signal according to the present invention in the band limiting block 230 is illustrated in FIG. 9, which shows a process of performing filtering at the position of f (3,3). As shown. In FIG. 9, the result of the filtering is obtained by adding up the product between the filter coefficient and the overlapping pixels. If the two-dimensional low pass filtered signal at the position of (x, y) is Z (x, y), this value is calculated by the following equation.

[Equation 3]

In Equation 3, since T means the order of the filter, T = 3. Thus, u and v values have integer values between -1 and 1. Also, in Equation 3, k (u, v) is a filter coefficient value, and f (x, y) is a pixel value. At this time, as in the above-described one-dimensional low-pass filtering, if the (xu), (yu) value of f (xu, yv) in the above formula is less than 0, it is set to 0, or corresponds to the image size of one frame If it becomes larger than the M-1 value, the M-1 value is used. Therefore, if the filtering is performed at all pixel positions as in the above equation, as shown in FIG. 9, two-dimensional filtering in the horizontal and vertical directions may be obtained.

On the other hand, when the band limiting technique using the DFT among the techniques for removing the high frequency components of the reconstructed frame, the band limiting block 230 is based on the filtering control signal B provided from the control block 220 described above. By limiting the high frequency component which is relatively insensitive to the visual in the reconstructed frame, the process can be divided into two-dimensional frequency conversion process and frequency selection process. In this process, the discrete Fourier transform (DFT) is used in the two-dimensional frequency conversion process. In the frequency selection process, the passband of the DFT-converted video signal is determined based on the filtering control signal B provided from the control block 220.

Next, a process of selecting the frequency of the image signal subjected to the two-dimensional DFT conversion based on the filtering control signal B and performing the two-dimensional DFT conversion on the frame reconstructed by the band limiting block 230 will be described in detail.

First, the band limiting block 230 uses the similarity of the spatial domain of the reconstructed video signal. The band-limiting block 230 uses the Fourier function to convert the spatial video signal (pixel data) according to the following equation. For example, it is transformed into quadratic DFT transform coefficients in a frequency domain of 8x8 units.

[Equation 4]

In Equation 4, f (u, v) denotes a value of each pixel, u denotes a position in a horizontal direction, and v denotes a position of a pixel in a vertical direction. Therefore, the value for each pixel of the N × N block has the following value. That is, when N = 8 u and v have a value between 0 and 7. At this time, each value has an integer value between 0 and 255. In addition, in Equation 6, Z (k, l) means a converted signal, k, l means a frequency component in the horizontal and vertical direction, respectively. Therefore, when N = 8, an 8x8 DFT block, that is, a signal in the spatial domain is converted into a signal in the frequency domain.

More specifically, in the band limiting block 230, the DFT conversion coefficients obtained through the above-described process are passed based on the filtering control signal B provided from the control block 220 through the line L13. Determine the frequency band. In this case, as shown in Equation 5, the filtering control signal B may be set to an integer value between 1 and 4, and the selected frequency is as follows.

That is, the band limiting block 230 selects a specific frequency from the converted frequency Z (k, l). Here, k and l are integer values between 0 and N-1. Thus, the value output from the band limit block 230 is a signal from which a specific frequency component (ie, a high frequency component) has been removed. For example, when N = 8, as shown in FIG. 4 as an example, the pass frequency band will be determined.

Therefore, as shown in FIG. 4, the frequencies below the dotted line corresponding to each of them are selected to be 0 according to the filtering control signal B provided from the control block 220 to the band limiting block 230 through the line L13. I never do that. That is, when the B value is 4 in FIG. 4, frequencies below the dotted line such as Z (1, 7), Z (2, 6), etc. are all mapped to 0. FIG. On the contrary, in response to the filtering control signal B from the control block 220 using one preset filter coefficient, the high frequency component of the predetermined level or more is replaced (such as replacing the high frequency component of the fixed level or more with 0). In this case, the implementation will be somewhat easier than the adaptive (or selective) band limit.

Next, as described above, the DFT transform coefficients (signals in the frequency domain) from which the frequency (high frequency component) of the specific region is removed according to the filtering control signal B value determined based on the complexity of the image are obtained by the following equation. The signal is inversely transformed into the spatial domain of.

[Equation 5]

In Equation 5, f (u, v) means the value of each pixel, u and v means the position of the pixel in the horizontal and vertical direction, Z (k, l) means the converted signal, k and l mean frequency components in the horizontal and vertical directions, respectively. Therefore, when N = 8, 8x8 DFT blocks in the frequency domain are converted into signals (pixel data) in the spatial domain.

As a result, in the band limiting block 230, when the reconstruction frame is a complex image on the line L17, it is calculated based on the complexity of the reconstructed image, in which the frequency of the specific region is selectively removed according to the complexity of the image. In response to the filtering control signal B, a reconstructed frame signal (video signal in which a high frequency component of a specific region is replaced with a value of 0) is selectively (or adaptively) removed, and the generated signal (human The reconstructed frame signal in which the high frequency component of which is relatively insensitive to the visual characteristic of?) May be provided to the D / A converter 108 of the next stage through the ba line of the switching block 240.

On the other hand, when the band limiting technique using DCT is adopted among the techniques for removing high frequency components of the recovered frame, the band limiting block 230 is based on the filtering control signal B provided from the control block 220 described above. In this reconstructed image, high frequency components are relatively insensitive to human vision, and the process can be divided into two-dimensional frequency conversion process and frequency selection process. In this process, discrete cosine transform (DCT) In the frequency selection process, the passband of the DCT-converted video signal is determined based on the filtering control signal B provided from the control block 220.

Next, a process of selecting a frequency of the 2D DCT transformed image signal based on the filtering control signal B for the frequency domain classification and reconstructing the image reconstructed by the band limiting block 230 is attached. It demonstrates in detail with reference to FIG.

First, in the band limiting block 230, two-dimensional DCT conversion is performed on a reconstructed image provided from the frame memory 107. As is well known in the art, the DCT conversion process is performed in a spatial manner of an image signal. It is known to reflect the similarity well, and this DCT converter method is widely applied in the process of encoding a video signal. Therefore, detailed description is omitted here. Therefore, in the present embodiment, when the reconstructed image is a complex image, DCT transform having such characteristics (reflecting spatial similarity) is used as an effective frequency selection technique according to the complexity of the image signal. The frequency selection process according to the present invention returns to the frequency domain reflecting the characteristics of the image signal better than the simple frequency converter method, and as a result, it is possible to increase the efficiency when selecting a frequency for the input image.

FIG. 10 shows a detailed block diagram of the band limiting block 230 according to the present invention shown in FIG. 2. As shown in the figure, the band limiting block 230 includes a DCT block 1141, a quantization block 1143, a frequency selector 1145, an inverse quantization block 1147, and an IDCT block 1149.

In FIG. 10, the DCT block 1141 uses the similarity of the spatial domain of the video signal. The DCT block 1141 uses the cosine function to convert the video signal (pixel data) of the spatial domain according to the following equation. For example, two-dimensional DCT transform coefficients of a frequency domain of 8x8 units are converted and provided to the next quantization block 1143.

[Equation 6]

In Equation 6, F (u, v) means a transformed DCT coefficient, and f (x, y) means an input video signal. Here, x and y denote positions in the horizontal and vertical directions of the pixel data, and u and v denote frequencies in the horizontal and vertical directions in the converted DCT coefficients. The quantization block 1143 then quantizes the DCT coefficients two-dimensionally transformed by Equation 6 to a finite number of values through, for example, nonlinear operations. In this case, the QP value is used in the quantization process of the DCT transform coefficient. When the transformed DCT coefficient is F (u, v), an operation of performing an integer value by performing F (u, v) / (2 * QP) is performed. This is an example of a representative quantization process.

Meanwhile, in the frequency selector 1145, the DCT transform coefficients quantized through the above-described process are applied to the filtering control signal B for determining the frequency bandwidth provided from the control block 220 of FIG. 2 through the line L13. Based on that, determine the frequency passed. In this case, as shown in Equation 1, the filtering control signal B for bandwidth determination may be set to an integer value between 1 and 4, and the selected frequency is as follows.

That is, the frequency selector 1145 selects a specific frequency from the converted frequency Z (k, l), where k, l is an integer value between 0 and N-1. Therefore, the value output from the frequency selector 1145 becomes a signal from which a specific frequency component (ie, a high frequency component) is removed. For example, if N = 8, its pass frequency will be determined as shown in FIG.

Thus, as shown in FIG. 4, the frequencies below the dotted line corresponding to each of the zeros according to the filtering control signal B for determining the bandwidth provided from the control block 220 to the frequency selector 1145 through the line L13 are all zeros. Do not select That is, when the B value is 4 in FIG. 4, frequencies below the dotted line such as Z (1,7), Z (2,6), etc., are all mapped to 0. FIG. On the contrary, in response to the filtering control signal B from the control block 1120 using one preset filter coefficient, the high frequency component of a predetermined level or more may be limited (such as replacing 0 with a high frequency component of a fixed level). It may be, in which case the implementation will be somewhat easier compared to the adaptive (or optional) band limitation.

Next, as described above, when the reconstructed image provided from the frame memory 107 in the frame is an image having a large complexity, the filtering control signal B value for determining the bandwidth determined based on the complexity of the reconstructed image is determined. Accordingly, the quantized DCT transform coefficients from which the frequency (high frequency component) of the specific region is removed are restored to the original signal (pixel data) through the inverse quantization block 1147 and the IDCT block 1149 of the next stage. At this time, the inverse transformation process of the inverse quantized DCT transformation coefficient in the IDCT block 1149 is as shown in the following equation.

[Equation 7]

In Equation 7, f (x, y) denotes an inversely transformed video signal (pixel data), and F (u, v) denotes a transformed DCT coefficient. Here, u and v denote horizontal and vertical frequencies in the converted DCT coefficients, and x and y denote horizontal and vertical positions of the pixel data.

As a result, in the IDCT block 1149, the reconstructed image is calculated based on the complexity of the reconstructed image, in which the reconstructed image is an image having a large complexity and the frequency of a specific region is selectively removed according to the complexity of the image. According to the filtering control signal B for bandwidth determination, a reconstructed frame signal (reconstructed image signal in which a high frequency component of a specific region is replaced with a zero value) that is selectively (or adaptively) removed from a high frequency component of an image is generated. The generated signal (a reconstruction frame signal in which the high frequency component relatively insensitive to the human visual characteristic is selectively removed) will be provided to the D / A converter 108 of the next stage through the ba line of the switching block 240.

Therefore, the D / A converter 108 of FIG. 2 recovers when employing a filtering technique (one-dimensional or two-dimensional low pass filtering, band limitation using DFT or DCT) for high frequency component removal according to the present invention. If the image is a complex image, the band limiting block 230 converts a reconstructed frame signal from which a high frequency component relatively insensitive to human visual characteristics is removed to an analog signal, and provides it to a display, not shown, and the reconstructed image is a simple image. By providing a reconstructed frame signal from which the high frequency component outputted from the back frame memory 107 is not removed to the display side, it is possible to reproduce the low frequency signal, which is a visually important component even with a high complexity image, with less quantization error. It is.

As described above, according to the present invention, it is determined whether a reconstructed frame is a complex image based on quantization parameter information in units of macroblocks in the reconstructed frame signal, and as a result, a filtering technique is performed on the subsequent reconstructed frame signal which is continuous in time. By removing the high frequency components insensitive to human temporal characteristics, it is possible to effectively prevent deterioration of image quality due to quantization error inevitably seen in the reproduced image even when the reconstructed image has a large complexity.

Claims (23)

Image reconstruction means for reconstructing the original signal before encoding by applying variable length decoding, inverse quantization, inverse DCT, and motion compensation techniques to the coded image bitstream, and digitally converting the reconstructed frame signal into an analog signal and providing it to the display side. And an analog conversion means, wherein the inverse quantization is performed by using quantization parameter of each macro block unit received together with the video bit stream. An average quantization parameter calculation block for averaging a plurality of quantization parameter values in macroblock units for a currently reconstructed frame to calculate an average quantization parameter value of a reconstructed frame; The complexity of the currently reconstructed frame is calculated by comparing the average quantization parameter value of the calculated one frame with a preset reference value, and immediately based on the calculated complexity information, in time rather than the frame used for calculating the complexity. A control block for generating a filtering control signal for the restored next frame which exists later; Band limiting means for generating a reconstructed frame from which high frequency components are removed by filtering the reconstructed frame based on the filter coefficient determined according to the generated filtering control signal and limiting the pass band; And At least two paths for providing the reconstructed frame signal to the analog conversion means, and based on a comparison result of an average quantization parameter value for the calculated one frame provided from the control block with a preset reference value. And a switching block for providing the reconstructed frame signal to the digital / analog converting means or the reconstructed frame signal from which the high frequency component is removed to the digital / analog converting means in response to a switching control signal generated. Video decoding system. The method of claim 1, The predetermined reference value is an improved image decoding system, characterized in that the intermediate value of the integer unit quantization parameter values in the preset category. The method of claim 1, And said band limiting means removes high frequency components of said reconstructed next frame signal through one-dimensional low pass filtering. The method of claim 3, The one-dimensional low pass filtering is an improved image decoding system, characterized in that to completely or partially remove the high frequency components of the reconstructed next frame signal using a predetermined filter coefficient. The method of claim 3, And the one-dimensional low pass filtering selectively removes the high frequency component level of the restored next frame using any one of a plurality of filter coefficients having different values. The method according to claim 3, 4 or 5, The 1D low pass filtering is improved image decoding system, characterized in that performed in units of 8 × 8 blocks. The method of claim 6, 1D low pass filtering on the 8x8 block is performed sequentially in the horizontal-vertical or vertical-horizontal direction of each pixel position within the 8x8 block. The method of claim 1, And said band limiting means removes high frequency components of said reconstructed next frame through two-dimensional low pass filtering. The method of claim 8, The two-dimensional low pass filtering is an improved image decoding system, characterized in that the complete high-frequency component of the reconstructed next frame using one filter coefficient. The method of claim 8, And the two-dimensional low pass filtering selectively removes the high frequency component level of the restored next frame using any one of a plurality of filter coefficients having different values. The method of claim 10, Wherein each filter coefficient has an order of 3x3. The method according to any one of claims 8 to 11, 2D low pass filtering on the reconstructed next frame is performed in units of 8 × 8 blocks. The method of claim 1, The band limiting means converts the image signal in the spatial domain for the reconstructed next frame into DFT transform coefficients in the frequency domain in M × N block units using a Discrete Fourier Transform, and filters for the generated band limitation. Determine a high frequency pass band for the transformed DFT transform coefficient blocks based on a control signal, limit the high pass band of each transformed DFT transform coefficient block to the determined bandwidth, and each of the DFT blocks for which the bandwidth is limited And a reconstructed frame signal from which the high frequency component has been removed by reconstructing the original signal through an inverse discrete Fourier transform for each. The method of claim 13, The Discrete Fourier Transform for the restored next frame is performed in units of 8 × 8 blocks. The method of claim 13, The bandwidth limitation of the restored next frame is adaptively performed to any one of a plurality of levels preset according to the complexity of the input frame. The method of claim 13, The bandwidth limitation of the restored next frame is performed at a level within a predetermined range. The method according to any one of claims 13 to 16, The bandwidth limitation is improved image coding system having an adaptive encoding mode determination function, characterized in that for each of the transformed DFT transform coefficient blocks, a high frequency component below the determined bandwidth is mapped to a zero value. The method of claim 1, The band limiting means is: Discrete cosine transform means for converting an image signal in a spatial domain with respect to the reconstructed frame signal into two-dimensional DCT transform coefficients in a frequency domain in units of M × N blocks using a cosine function; Quantization means for quantizing the two-dimensional DCT transform coefficient blocks in M × N units using a quantization parameter value to a finite number of values; Determining a high frequency pass band for the quantized DCT transform coefficient blocks based on the filtering control signal provided from the control block, and limiting a high frequency pass band of each quantized DCT transform coefficient block to the determined bandwidth. Frequency selection means; And Inverse quantization and inverse discrete cosine transform are performed on each of the bandwidth-limited quantized DCT blocks to reconstruct the original signal before encoding to generate a band-limited frame, and the band-limited frame is a reconstructed frame from which the high frequency component is removed. And an image reconstruction means for providing the switching block as a signal. The method of claim 18, The discrete cosine transform on the reconstructed next frame is performed in units of 8 × 8 blocks. The method of claim 18, The bandwidth limitation of the restored next frame is adaptively performed to any one of a plurality of levels preset according to the complexity of the recovered frame. The method of claim 18, The bandwidth limitation of the restored next frame is performed at a level within a predetermined range. The method according to any one of claims 18 to 21, The bandwidth limitation is an improved image decoding system, characterized in that for the transformed each DCT transform coefficient block to map a high frequency component below the determined bandwidth to a zero value. The method of claim 1, And said switching control signal is a logic signal having a high or low level.