KR100229533B1 - Improved image decoding system - Google Patents

Abstract

According to the present invention, the complexity of an image reconstructed into an original signal is calculated on the basis of short-time statistics of a plurality of frames for a predetermined period of time using the standard deviation of each reconstructed frame, The present invention relates to an image decoding system suitable for suppressing image deterioration due to a quantization error by adaptively removing a high frequency component of a signal, A statistical characteristic calculating block for extracting a standard deviation using an average value of pixels representing statistical characteristics through a statistical characteristic calculating block; Storing a plurality of standard deviation values for each frame for a predetermined predetermined time that has been previously restored, provided from the statistical characteristic calculation block, calculating an average value of the stored plurality of standard deviation values, A control block for calculating a complexity of a currently reconstructed frame through comparison with a standard deviation value of the reconstructed frame and generating a filtering control signal for a frame reconstructed based on the calculated complexity; A band limitation unit for filtering a currently reconstructed frame based on a filter coefficient determined in accordance with the generated filtering control signal and limiting a pass band thereof to generate a reconstructed frame from which a high frequency component is removed; And at least two paths for providing the reconstructed frame signal to the analog conversion means, wherein the response to the switching control signal generated based on the comparison result between the average value provided from the control block and the standard deviation value of the currently reconstructed frame And a switching block for providing the reconstructed frame signal to the digital / analog conversion means or providing the reconstructed frame signal from which the high-frequency component has been removed to the digital / analog conversion means. Thus, image quality deterioration due to quantization error Can be effectively reduced.

Description

Improved Image Decoding System

The present invention relates to a video decoding system for decoding an encoded video signal, and more particularly, to a video decoding system for decoding a video signal having a standard deviation (dispersion value) for a plurality of restored frame signals for a predetermined time The present invention relates to an improved image decoding system suitable for detecting a complexity of an image and compensating for deterioration in image quality caused by a quantization error after decoding based on the detected complexity.

MPEG-1 and MPEG-2 (the official name of this recommendation is ITU-T Rec.H222.0 | ISO / IEC 13818) are widely known as general methods for decoding compression-encoded video signals. In such a recommendation, when a video signal is decoded, an input video stream conforms to a format defined by MPEG, so that a device for decoding in accordance with this format may be designed.

However, since a general decryption method is recommended in this process, an efficient method and apparatus for decoding the decryption are still actively studied all over the place.

1 is a block diagram of a typical image decoding system. As shown in the figure, the bit stream of the encoded video signal separated and input through the system decoding system is input to and stored in the reception buffer 101. The bit stream stored in the reception buffer 101 is supplied to the variable length decoding unit 102 at predetermined time intervals in synchronization with the decoding timing and is subjected to a variable length decoding process which is an inverse process of variable length coding in the transmission side image coding system .

(Quantized DCT transform coefficients) and motion vectors (motion displacements) obtained therefrom are separated from each other, and the separated quantized streams (quantized DCT transform coefficients) are supplied to the next-stage inverse quantizer 103 And the motion vectors are provided to the motion compensation unit 104, which will be described later, for motion compensation.

Then, the DCT transform coefficients, which are inversely quantized through the inverse quantization unit 103, are provided to the inverse DCT unit 105. The inverse DCT unit 105 performs inverse DCT on the DCT transform coefficients using an inverse process of the DCT process in the transmission side image encoding system, that is, the inverse discrete cosine transform, A difference signal using motion estimation and compensation using the previous frame). That is, the reconstructed signal of the original signal output from the inverse DCT unit 105 is a difference signal between the current frame and the current frame and the prediction frame obtained through motion compensation / compensation between the current frame and the previous frame.

Therefore, in the video decoding system, the differential signal from the inverse DCT unit 105 and the motion compensated frame signal provided from the motion compensation unit 104 (that is, supplied from the frame memory 107 using the motion vectors) The frame reconstructed from the reconstructed previous frame) is added through the adder 106 to generate a reconstructed image frame as the original signal.

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

On the other hand, the motion compensation unit 104 compensates for the motion of the current frame from the previous frame based on the previous frame stored in the frame memory 107 and the motion vector (or displacement) of the variable length decoding unit 102, And provides the reconstructed frame signal as a motion compensated frame signal to the adder 106. The adder 106 adds the reconstructed frame signal to the reconstructed frame signal.

Therefore, the adder 106 adds the motion compensated image frame signal from the motion compensating unit 104 and the difference signal from the inverse DCT unit 105 to generate a continuous reconstructed frame signal, And supplies it to the memory 107.

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

The encoded bitstream received through the reception channel and stored in the reception buffer 101 is decoded at a decoding time point. The decoded time point is determined by extracting a specific parameter transmitted from the transmission side image encoding system , The decryption time point is related to the size of the reception buffer 101 and it is determined that a malfunction (overflow, underfolw) 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 input at the same transmission rate in time, a malfunction of the receiving buffer 101 will not occur.

However, the amount of bits generated at the time of encoding in the transmission side image encoding system is different for each frame. That is, when the image is very complex, a relatively large amount of bits is generated after the encoding, and when the image is simple, a relatively small amount of bits are generated after the encoding. For this reason, in the image encoding system, considering the capacity and the transmission rate of the output- By adjusting the quantization step size, the bit generation amount is appropriately adjusted.

In other words, when the image to be encoded is a relatively complex image in the transmission side encoding system, the amount of bits after encoding becomes large. In the image encoding system, the size of the quantization step size Limiting the bit generation amount by increasing the quantization parameter (QP) value. On the other hand, if the image to be encoded is a relatively simple image in the transmission-side encoding system, the amount of bits after encoding becomes small. In this case, in the image encoding system, by decreasing the quantization parameter (QP) do.

On the other hand, when an image signal quantized according to a quantization parameter determined in consideration of a bit generation amount and a transmission rate due to image complexity is quantized into a quantization parameter of a large value as described above, Deterioration in image quality, that is, deterioration in image quality due to blocking phenomenon that is conspicuous when the screen is finely divided is caused. In particular, such a quantization error mainly appears in a high frequency component relatively insensitive to human visual characteristics. If such an image is reproduced as it is, it is inevitable that an image with deteriorated image quality will inevitably be seen. Therefore, it is possible to reproduce an image having a more natural image quality if quantization error mainly distributed in a high frequency component can be removed.

Accordingly, the present invention has been made in view of the above points, and it is an object of the present invention to provide a method and apparatus for calculating a complexity of an image reconstructed from an original signal based on short-time statistics of a plurality of frames during a previous predetermined time using a standard deviation of each reconstructed frame, And an object of the present invention is to provide an improved image decoding system capable of suppressing deterioration of image quality due to quantization error by adaptively removing high frequency components of a video signal reconstructed according to a result of the calculation.

According to an aspect of the present invention, there is provided an image decoding apparatus including an image restoration unit for restoring an original image signal before coding by applying a variable length decoding, an inverse quantization, an inverse DCT, and a motion compensation technique to an encoded image bitstream, And a digital-to-analog conversion means for converting the digital image signal into a digital image signal and providing the digital image signal to a display side, wherein the average value of the pixels representing the statistical characteristics is used for the reconstructed frames provided from the image reconstructing means A statistical characteristic calculation block for extracting a standard deviation; Storing a plurality of standard deviation values for each frame for a predetermined period of time previously restored provided from the statistical characteristic calculation block, calculating an average value of the stored plurality of standard deviation values, A control block for calculating a complexity of the currently reconstructed frame through comparison with a standard deviation value of a currently reconstructed frame and generating a filtering control signal for the currently reconstructed frame based on the calculated complexity; A band limitation unit for filtering the currently recovered frame based on a filter coefficient determined according to the generated filtering control signal and limiting a pass band thereof to generate a reconstructed frame from which a high frequency component is removed; And at least two paths for providing the reconstructed frame signal to the analog conversion means, wherein the at least two paths are generated based on a result of comparison between the average value provided from the control block and the standard deviation value of the currently reconstructed frame Further comprising a switching block for providing the reconstructed frame signal to the digital / analog conversion means in response to a switching control signal or for providing a reconstructed frame signal from which the high-frequency component has been removed to the digital / analog conversion means, System.

FIG. 1 is a block diagram of a conventional image decoding system,

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

FIG. 3 is a diagram showing a decision area for limiting a high frequency component, which is determined as one fixed level with respect to an 8 × 8 pixel block, as an example when the reconstructed image according to the present invention is a complex image;

FIG. 4 is a diagram illustrating a decision area for a high-frequency component restriction that is adaptively determined based on the complexity of an 8 × 8 pixel block as an example when the reconstructed image is an image having a high complexity,

FIG. 5 is an illustration of an 8x8 pixel block as an example according to the present invention,

FIG. 6 is an exemplary diagram illustrating a one-dimensional low-pass filter coefficient having an order of 7 as an example according to the present invention,

FIG. 7 is an exemplary diagram illustrating horizontal filtering at a (0, 4) position and vertical filtering at a (3, 0) position according to another embodiment of the present invention,

FIG. 8 is an example of a 3 × 3-order two-dimensional low-pass filter coefficient according to another embodiment of the present invention,

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

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

Description of the Related Art

101: Receive buffer 102: Variable length decoder

103: Inverse quantization unit 104: Motion compensation unit

105: inverse DCT 106: adder

107: frame memory 108: D / A converter

210: statistical characteristic calculation block 220: control block

230: Band limiting block 240: Switching block

1141: DCT block 1143: Quantization block

1145: Frequency selector 1147: Inverse quantization block

1149: IDCT block

These and other objects and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.

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

In the present invention, when it is judged that the video signal reconstructed by the original signal before coding through the variable length decoding, inverse quantization and inverse DCT is a complex video or a simple video, In order to prevent deterioration of image quality due to a quantization error by adaptively removing high frequency components relatively insensitive to human visual characteristics in a video signal, the complexity of the reconstructed video signal is calculated by calculating the spatial complexity of the image, The spatial complexity of the frames is obtained by using the short-term statistics of a plurality of frames for a predetermined time based on the average and standard deviation of pixel values of each frame and the average and standard deviation of each frame reconstructed.

Generally, the spatial complexity of a video signal is related to the amount of information of a video signal to be encoded. Such complexity can be calculated by various methods. One of the preferable methods is, for example, the use of a variance value (standard deviation) . This is because, if the variance value is large, the result of performing the DCT will include many high-frequency components, so that the distribution of the coefficient of variation becomes inadequate to compress the data.

Ideally, a good image for compression is a case where there is no high frequency component but only a DCT component, and the distribution of the transformed coefficients is only one value at the position of (0, 0). If the variance value is large, the motion compensation may not be performed properly and the structure of the motion compensated image is not suitable for coding. Therefore, a quantization error is inevitably generated in a process of encoding an image signal having a large spatial complexity, and conversely, a quantization error is relatively small in an image signal having a small spatial complexity. Therefore, considering this point, it can be said that the complexity of the reconstructed image signal is sufficient to consider the complexity of the image at the time of coding. That is, the complexity of the video signal at the time of encoding can be sufficiently predicted by using the complexity of the reconstructed video signal output from the decoding system.

Therefore, in the present invention, the complexity of the image is detected using the complexity of the reconstructed image signal, and the image is restored using the detection result to perform bandwidth limitation of the image provided to the display side, that is, We want to adaptively remove the components. That is, high-frequency components are adaptively removed for complex images with large quantization errors, and band limitation is not applied for images with low complexity where quantization errors are small.

2 shows a block diagram of an improved image decoding system according to a preferred embodiment of the present invention. The video decoding system of the present invention shown in the figure includes a reception buffer 101, a variable length decoding unit 102, an inverse quantization unit 103, a motion compensation unit 104, an inverse DCT unit 105, an adder 106, And a frame memory 107, a D / A converter 108, a statistical characteristic calculating block 210, a control block 220, a band limiting block 230 and a switching block 240.

2, the image decoding system of the present invention includes a statistical characteristic calculating block 210, a control block 220, a band limiting block 230, And a switching block 240. The additional constituent elements of the present invention have an object of the present invention, that is, an object of the present invention is to provide an image quality enhancement apparatus, The object of suppressing deterioration can be achieved.

Therefore, as described above, the remaining constituent members except for the constituent members 210, 220, 230, and 240 added to the conventional decoding system represent the same constituent members that perform substantially the same functions as in the conventional decoding system, so in order to avoid unnecessary redundant description, The detailed description thereof will be omitted.

Referring to FIG. 2, the statistical characteristic calculating block 210 receives a frame signal restored from the frame memory 107 through a line L11, and calculates a parameter capable of determining whether the image has a large complexity In the present invention, the average MIp and the standard deviation ACT of the video signal are used as these parameters. That is, as an example, if the image reconstructed immediately before is Ip and the current reconstructed image is Ic, the pixel value of the Ip image at the position (x, y) is Ip (x, y). In this case, the computation of the complexity of the currently reconstructed Ic image can be performed using the ACT calculated using the previously reconstructed Ip image as shown in the following equation.

In Equations (1) and (2), M and N are integer values indicating the horizontal and vertical sizes of the video signal. The thus extracted parameters, i.e., the standard deviation ACT of each frame, are provided to the next stage control block 1120. At this time, if the ACT value is large, it means that the image is a complex image. Conversely, if the ACT value is small, it means that it is a simple image.

The reason why the standard deviation ACT extracted as described above is used as the statistical characteristic information to be used in determining whether the reconstructed image is a complex image is that if the reconstructed image signal is not an image having a large complexity, The statistical characteristics (standard deviation) of the current restored video signal are not significantly different from those of the previously restored video signal. On the other hand, if the currently restored video signal is a video having a large complexity, This is because it is very different from the statistical characteristics of video signals. Accordingly, in the present invention, it is determined whether or not the current reconstructed image is complex using the temporal change of the statistical characteristic value, and the determination will be performed in the control block 220 described below.

In the present invention, considering the fact that the video signal does not change abruptly every frame, the previously reconstructed image is used in calculating the complexity of the reconstructed image. This is because the computation This is to consider the processing time. That is, if the currently reconstructed image data is used after the current reconstruction process is completely performed, a delay time of one frame is generated. However, if the complexity of the reconstructed image is used as the complexity of the reconstructed image, The band limiting process according to the present invention can be performed in the image restoration process.

Meanwhile, in the control block 220, using the parameters provided from the statistical characteristic calculation block 210, i.e., the standard deviation (ACT) values of each reconstructed frame, And the filtering control signal B corresponding to the determination result is provided to the band limitation block 230 via the line L 13.

In more detail, in order to detect whether the reconstructed image is a complicated image, the control block 220 calculates a statistical characteristic (a short time statistic for a previously reconstructed frame) for a predetermined time and a current statistical characteristic The parameters (standard deviation) of the restored frame are compared with each other to judge whether or not the complexity is present.

In the present invention, the standard deviation of each frame with respect to the previously recovered frames of one second is used. However, in the present invention, . In the control block 220, an ACT value for each previously recovered frame provided from the statistical characteristic calculation block 210 is input, and a standard deviation for one second is calculated for these values. For this calculation, the control block 220 stores the ACT value input every frame in the statistical characteristic calculating block 210 in an embedded memory area (not shown).

That is, the parameter extracted from the reconstructed image (frame) is ACT (0), the parameters extracted from the reconstructed image are ACT (1), and ACT (i) , The parameter value input 30 frames before is ACT 30, and these values are stored in the memory area. Accordingly, the control block 220 calculates the average value M of these values from the 30 values from the standard deviation ACT (1) to the ACT 30 of each restored frame to obtain the short-term statistics (I.e., a statistical characteristic of the product). Of course, the oldest ACT value among the 30 ACT values input from the past statistical characteristic calculation block 210 is discarded.

Accordingly, if it is determined in the control block 220 that the ACT value of the current restored frame is a frame having a more complex image than the statistical characteristic (the average value of each frame ACT for the restored 1 second) M, A filtering control signal B is generated to control the filtering operation in the band limiting block 230 and a switching control signal (logic signal having a high level or a low level) is generated on the line L15, Controls the switching operation, that is, when the contact is connected to ac, the contact is connected to ab.

On the other hand, in the band limitation block 230, adaptive band limitation processing, that is, filtering is performed on the currently reconstructed frame signal provided from the frame memory 107 based on the filtering control signal B on the line L13 to generate a predetermined high- And generates the removed restoration frame signal on the line L17. That is, when filtering the restored frame signal, the band limitation block 230 may set all values of Z (1,7) and Z (2,6) And the bandwidth is limited and filtering is performed.

Then, the restored frame signal from which the high-frequency component has been removed is output to the D / A converter 108 at the next stage via the contact b-a of the switching block 240. Therefore, the D / A converter 108 converts the restored frame signal from which the high-frequency component has been removed into an analog signal, and then provides it to a display (not shown) for display on a monitor.

On the other hand, in the control block 220, when the band limitation block 230 filters the reconstructed frame signal provided in the frame memory 107 to generate a reconstructed frame signal from which high-frequency components have been removed, The filtering control signal B can be calculated according to the following equation so that it can be adaptively (or selectively) filtered in consideration of the complexity of the input image and the like without filtering the reconstructed frame at a level (i.e., a filter coefficient)

As a result, in the band limitation block 230, an appropriate band limitation process corresponding to the restored frame signal provided from the frame memory 107, that is, adaptive filtering is performed based on the filtering control signal B on the line L13 . That is, when the restored frame signal is filtered, the band limitation block 230 corresponds to the filtering control signal B from the control block 220 based on the complexity of the input image, for example, as shown in FIG. , The value of the output value or less is mapped to a value of zero to limit the bandwidth, that is, to perform filtering. In other words, according to the present embodiment, the high frequency component removal level of the reconstructed frame signal is selectively (or adaptively) adjusted according to the complexity of the reconstructed frame.

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

If it is determined that the inputted AQP value is larger than the reference value, the control block 220 generates a high-level switching control signal on the line L15, . Accordingly, the restored frame signal from which the high-frequency component has been removed through the band limiting block 230 is output to the D / A converter 108 at the next stage.

On the other hand, according to the present invention, when the band limitation block 230 removes a high frequency component from a frame signal reconstructed through filtering, a one-dimensional low-pass filtering technique, a two-dimensional low- Technique, a band limitation technique using a discrete Fourier transform (abbreviated as DFT), and a band limiting technique using a discrete cosine transform (abbreviated as DCT).

As described above, according to the present invention, the one-dimensional low-pass filtering technique, which is one of techniques for eliminating high-frequency components relatively insensitive to human visual characteristics, is performed by multiplying an input restored frame (image) signal by a low- do. That is, assuming that the value of each pixel for an image of N × M blocks (for example, 8 × 8 blocks) is f (x, y), for example, In the case of a block of x 8, the position values x and y in the horizontal and vertical directions of the pixel 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 horizontal and vertical position values of each pixel of the 8x8 block have a value of f (7,7) at f (0, 0).

On the other hand, as the one-dimensional low-pass filter in the present invention, for example, as shown in Fig. 6, it is assumed that the one-dimensional low-pass filter coefficient has seven orders. In this low-pass filter, when the sampling frequency of the input image signal is fs, the frequency bandwidth is fs / 2, and therefore, the low-pass filter is a band-pass filter that limits the frequency band to have a frequency bandwidth of fs / 4.

Accordingly, in the process of performing the one-dimensional low-pass filtering on the frame signal reconstructed according to the present invention in the band limitation block 230, since the reconstructed frame signal is a two-dimensional signal in the horizontal and vertical directions, And performing filtering. This process is illustrated in detail in FIG. 7 which illustrates the horizontal filtering at position (0.4) and the vertical filtering at (3,0) position. That is, if a signal that is low-pass-filtered in the vertical direction at the position of (x, y) is z (x, y), this is calculated by the following equation.

In Equation (4), T means the order of the filter, so T = 7. Thus, the u value has an integer value between -3 and 3. In Equation (4), k (u) is a filter coefficient value and f (x, y) is a pixel value. If the (yu) value of f (x, yu) in the above equation (4) is smaller than 0 or is larger than the M-1 value corresponding to the image size of one frame as a whole, As shown in Fig. This is a method of setting a pixel value only in the region, that is, in the range of 0 to M-1, so that it is set as an end value in the case of exceeding this region. Such a filtering method is a technique well known in the art.

Accordingly, when filtering is performed at all pixel positions as in Equation (4), one-dimensional filtering in horizontal and vertical directions can be obtained as shown in FIG. 7, for example. At this time, in the process of performing such filtering at the positions of all the pixels according to the present invention, the horizontal direction filtering may be performed first, and the vertical direction filtering may be performed again on the horizontal direction filtered results or the order may be changed.

On the other hand, the high-frequency component of the reconstructed frame signal may be completely removed or partially removed using only one filter coefficient preset (or fixed) in the one-dimensional low-pass filtering as described above. However, Frequency components of the reconstructed image may be adaptively (or alternatively) removed using a plurality of filter coefficients (e.g., four, etc.) according to the complexity of the complexity. In this case, when the reconstructed frame is adaptively filtered using a plurality of filter coefficients, the implementation of the hardware may be somewhat complicated as compared with the image filtering using a predetermined filter coefficient, but a high- Is possible. Therefore, when adaptive filtering is performed using a plurality of filter coefficients as described above, it can be selectively applied according to the application range.

On the other hand, the two-dimensional low-pass filtering among the techniques for removing the high-frequency component of the reconstructed frame is performed on an image of one N × M block (for example, 8 × 8 block) Assuming that the value of each pixel is f (x, y) and, for example, the two-dimensional low-pass filter coefficient has a ninth order having three orders as shown in Fig. 8, K (0, 0) is 1/2, and the other values are 1/16.

Accordingly, the process of performing the two-dimensional low-pass filtering of the frame signal reconstructed according to the present invention in the band limitation block 230 may be performed in a manner similar to that shown in FIG. 9 Respectively. In Fig. 9, when the product between the filter coefficients and the pixels overlapping each other is added, a filtered result is obtained. If the two-dimensional low-pass filtered signal at the position (x, y) is Z (x, y), this value is calculated by the following equation.

In Equation (5), T denotes the degree of the filter, so T = 3. Therefore, u and v have an integer value between -1 and 1. In Equation (5), k (u, v) is a filter coefficient value and f (x, y) is a pixel value. At this time, as in the one-dimensional low-pass filtering described above, if the values of (xu) and (yv) of f (xu, yv) in the above formula are smaller than 0, If it is larger than the corresponding M-1 value, it is set to M-1. Accordingly, when filtering is performed at all pixel positions as in the above equation, for example, as shown in FIG. 9, two-dimensional filtering results in horizontal and vertical directions can be obtained.

On the other hand, in the case of adopting the band limitation technique using DFT among the techniques for removing high frequency components of the recovered frame, the band limitation block 230 may be configured to perform the filtering based on the filtering control signal B provided from the control block 220 In the reconstructed frame, a high frequency component relatively insensitive to time is limited. The process can be substantially divided into a two-dimensional frequency conversion process and a frequency selection process. In the two-dimensional frequency conversion process, a discrete Fourier transform (DFT) And determines the passband of the DFT-converted image signal based on the filtering control signal B provided from the control block 220 in the frequency selection process.

Next, a process of performing a two-dimensional DFT transform on the reconstructed frame in the band limitation block 230 and selecting a frequency of a two-dimensional DFT-transformed video signal based on the filtering control signal B will be described in detail.

First, the band limitation block 230 uses the similarity of the spatial domain of the reconstructed image signal. By using the Fourier function, the image signal (pixel data) in the spatial domain is multiplied by M × N units, Dimensional DFT transform coefficients in the frequency domain of, for example, 8x8 units.

In Equation (6), f (u, v) denotes the value of each pixel, u denotes the position in the horizontal direction, and v denotes the position of the pixel in the vertical direction. Therefore, the value for each pixel of the NxN block has the following values. That is, when N = 8, u and v have values between 0 and 7. At this time, each value has an integer value between 0 and 255. In Equation (6), Z (k, 1) denotes a converted signal, and k and l denote frequency components in the horizontal and vertical directions, respectively. Therefore, when N = 8, the 8x8 DFT block, that is, the signal in the spatial domain is converted into the signal in the frequency domain.

In more detail, in the band limitation block 230, on the basis of the filtering control signal B provided from the control block 220 through the line L13, the DFT transform coefficients obtained through the above- Band. At this time, the filtering control signal B can be set to an integer value between 1 and 4, as in Equation (3), and the frequencies selected are as follows.

That is, the band limitation block 230 selects a specific frequency from the converted frequency Z (k, l). Here, k and l are integers between 0 and N-1. Accordingly, the value output from the band limitation block 230 becomes a signal from which a specific frequency component (i.e., a high frequency component) is removed. For example, in the case of N = 8, the passing frequency band will be determined, as shown in Fig. 4 as an example.

4, according to the filtering control signal B provided from the control block 220 to the band limiting block 230 through the line L13, the frequencies below the corresponding dotted lines are all set to 0 I never do that. That is, in FIG. 4, when the B value is 4, frequencies below the dotted line such as Z (1, 7), Z (2, 6) Alternatively, one of the preset filter coefficients may be used to limit a high-frequency component of a predetermined level or higher (such as replacing a high-frequency component of a fixed level or higher with 0) in response to the filtering control signal B from the control block 220 Which may be somewhat easier to implement than an adaptive (or selective) bandlimit.

Next, the DFT transform coefficients (frequency domain signal) from which the frequency (high frequency component) of the specific region is removed according to the filtering control signal B value determined on the basis of the complexity of the image, as described above, As shown in FIG.

In Equation (7), f (u, v) denotes the value of each pixel, u and v denote the positions of pixels in the horizontal and vertical directions, Z (k, and kl represent frequency components in the horizontal and vertical directions, respectively. Therefore, in the case of N = 8, the 8x8 DFT blocks in the frequency domain are transformed into signals (pixel data) in the spatial domain.

As a result, in the band limitation block 230, when the restored frame has a complicated image on the line L17, the frequency of the specific region is selectively removed based on the complexity of the restored image, that is, (A video signal in which a high-frequency component of a specific region is replaced with a zero value) in which a high-frequency component of an image is selectively (or adaptively) removed in accordance with a filtering control signal B calculated by the filtering control signal B, (A reconstructed frame signal in which a high frequency component that is relatively insensitive to human visual characteristics is selectively removed) will be provided to the D / A changing portion 108 of the next stage through the ba line of the switching block 240.

On the other hand, when the DCT-based band limitation scheme is employed among the techniques for removing high-frequency components of the reconstructed frame, the band limiting block 230 performs a filtering operation on the basis of the filtering control signal B provided from the control block 220 In the reconstructed image, a high frequency component relatively insensitive to human vision is limited. The process can be substantially divided into a two-dimensional frequency conversion process and a frequency selection process. In the two-dimensional frequency conversion process, a discrete cosine transform And determines the passband of the DCT-converted video signal based on the filtering control signal B provided from the control block 220 in the frequency selection process.

Next, a process of two-dimensional DCT transforming the reconstructed image in the band limitation block 230 and selecting a frequency of the two-dimensional DCT-transformed image signal based on the filtering control signal B for frequency domain classification Will be described in detail with reference to FIG.

First, in the band limitation block 230, a 2D DCT transform is performed on a reconstructed image provided from the frame memory 107. As is well known in the art, the DCT transform process is a spatial transform This DCT transform technique is widely applied in the process of encoding an image signal. Therefore, detailed description here is omitted. In this embodiment, when the reconstructed image is a complex image, the 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 reflects the characteristics of a video signal better than a simple frequency conversion technique and returns to the frequency domain. As a result, the efficiency of the frequency selection for the input image can be increased.

FIG. 10 shows a detailed block diagram of the band limitation block 230 according to the present invention shown in FIG. The band limitation 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. By using the cosine function, the video signal (pixel data) in the spatial domain is multiplied by M × N units, Dimensional DCT transform coefficients in the frequency domain of, for example, 8x8 units, and provides the transform coefficients to the quantization block 1143 of the next stage.

In Equation (8), F (u, v) denotes a transformed DCT coefficient, and f (x, y) denotes an input image 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 transformed DCT coefficients. Then, the quantization block 1143 quantizes the two-dimensionally transformed DCT coefficients using Equation (8) into a finite number of values, for example, by nonlinear computation. In this case, if the transformed DCT coefficient is F (u, v), the operation of taking an integer value by performing F (u, v) / (2 * QP) is performed using the QP value in the quantization process of the DCT transform coefficients This is an example of a typical quantization process.

On the other hand, in the frequency selector 1145, the DCT transform coefficients quantized through the process described above are input to the filtering control signal B for determining the frequency bandwidth provided from the control block 220 of FIG. 2 through the line L13 And determines the frequency to be passed therethrough. At this time, the filtering control signal B for determining the bandwidth can be set to an integer value between 1 and 4, as shown in Equation (3).

That is, the frequency selector 1145 selects a specific frequency from the converted frequency Z (k, l). Here, k and l are integers between 0 and N-1. Therefore, the value output from the frequency selector 1145 becomes a signal from which a specific frequency component (i.e., a high frequency component) is removed. For example, if N = 8, the passband will be determined as shown in FIG.

Referring to FIG. 4, 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, frequencies below the corresponding dotted lines are all set to 0 Do not. That is, when the B value is 4 in FIG. 4, all frequencies below the dotted line such as Z (1, 7), Z (2, 6) Alternatively, one of the preset filter coefficients may be used to limit a high-frequency component of a predetermined level or higher (such as replacing a high-frequency component of a fixed level or higher with 0) in response to the filtering control signal B from the control block 1120 , Which would be somewhat easier to implement than an adaptive (or optional) bandlimit.

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

In Equation (9), f (x, y) denotes an inversely transformed image signal (pixel data), and F (u, v) denotes a transformed DCT coefficient. Here, u and v mean the frequencies in the horizontal and vertical directions in the transformed DCT coefficients, and x and y indicate positions in the horizontal and vertical directions of the pixel data.

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

Therefore, in the D / A conversion unit 108 of FIG. 2, when employing a filtering technique (one-dimensional or two-dimensional low-pass filtering, band limitation using DFT or DCT) for eliminating high frequency components according to the present invention, If the reconstructed image is a complex image, the reconstructed frame signal in which a high frequency component relatively insensitive to human visual characteristics is removed through a band limiting block 230 is converted into an analog signal and provided to a display (not shown) The reconstructed frame signal from which the high frequency component output from the frame memory 107 is not removed is provided to the display side so that even if the image has a large complexity, a visually important component can be reproduced with a low quantization error with respect to the low frequency signal. It is.

As described above, according to the present invention, the complexity of the presently reconstructed image is determined using the short-time statistics based on the standard deviation of a plurality of frames for a predetermined time that has been previously restored, By removing the high-frequency components insensitive to human visual characteristics using the filtering technique for the next reconstructed frame signal, even if the reconstructed image has a large complexity, the image quality deterioration due to the quantization error inevitably appears in the reconstructed image Can be effectively prevented.

Claims (11)

An image reconstruction means for reconstructing an original image signal before encoding by applying a variable length decoding, an inverse quantization, an inverse DCT, and a motion compensation technique to an encoded image bitstream, and a digital converting means for converting the reconstructed frame signal into an analog signal, / Analog converting means for calculating a statistical characteristic for extracting a standard deviation using an average value of pixels representing statistical characteristics through frequency analysis of each reconstructed frame provided from the image reconstructing means block; Storing a plurality of standard deviation values corresponding to each frame for a predetermined time that has been previously restored and provided from the statistical characteristic calculation block, calculating an average value of the stored plurality of standard deviation values, Calculating a complexity of the currently reconstructed frame by comparing the reconstructed frame with a standard deviation value of the currently reconstructed frame, and generating a filtering control signal for the currently reconstructed frame based on the calculated complexity; A band limiting unit for filtering the current recovered frame based on a filter coefficient determined according to the generated filtering control signal and limiting a pass band thereof to generate a recovered frame from which a high frequency component has been removed; And at least two paths for providing the reconstructed frame signal to the analog conversion means, wherein the control block is generated based on a result of comparison between an average value of the standard deviations and a standard deviation value of the currently reconstructed frame Further comprising a switching block for providing the reconstructed frame signal to the digital / analog conversion means in response to a switching control signal from the digital / analog conversion means, or for providing a reconstructed frame signal from which the high- Video decoding system. The improved video decoding system of claim 1, wherein the plurality of frames previously restored at predetermined predetermined times are reconstructed previous 30 frames immediately adjacent in time to the currently reconstructed frame. The apparatus of claim 1, wherein the band limiting unit selectively removes a high-frequency component level of the restored next frame through one-dimensional low-pass filtering using any one of a plurality of filter coefficients whose values are different from each other Wherein the image decoding apparatus further comprises: 4. The improved video decoding system of claim 3, wherein the one-dimensional low-pass filtering is performed sequentially in a horizontal-vertical or vertical-horizontal direction of each pixel position within the 8x8 block. The apparatus of claim 1, wherein the band limiting unit selectively removes a high-frequency component level of the restored next frame through two-dimensional low-pass filtering using any one of a plurality of filter coefficients whose values are different from each other Wherein the image decoding apparatus further comprises: The apparatus of claim 1, wherein the band limiting unit transforms the video signal of the spatial domain for the next frame restored into DFT transform coefficients in the frequency domain of MxN block using discrete Fourier transform, Determines a high-frequency passband for the transformed DFT transform coefficient blocks based on a filtering control signal for the band limitation, limits the high-frequency passband of each transformed DFT transform coefficient block to the determined bandwidth, And reconstructs the original signal through inverse discrete Fourier transform for each of the limited DFT blocks, thereby generating a reconstructed frame signal from which the high-frequency component has been removed. 7. The improved video decoding system of claim 6, wherein the bandwidth limitation of the restored next frame is adaptively performed at any one of a plurality of predetermined levels according to the complexity of the input frame. The adaptive encoding method according to claim 6 or 7, wherein the bandwidth limitation maps high frequency components of the determined bandwidth or less to each of the converted DFT transform coefficient blocks to zero (0) Improved image coding system with decision function. [2] The apparatus of claim 1, wherein the band limiting means comprises: a discrete cosine transform (DCT) transform for transforming a video signal in a spatial domain of the reconstructed frame signal into two-dimensional DCT transform coefficients in an M × N block unit frequency domain using a cosine function; Way; Quantization means for quantizing the 2-dimensional DCT transform coefficient blocks of M × N units into a finite number of values using a quantization parameter value; Frequency band for the quantized DCT transform coefficient blocks based on the filtering control signal provided from the control block and limits the high frequency passband of each quantized DCT transform coefficient block to the determined bandwidth Frequency selection means; And performing inverse quantization and inverse discrete cosine transform on each of the quantized DCT blocks with the limited bandwidth to reconstruct the original signal before coding to generate a band-limited frame, and restoring the band-limited frame by removing the high- And the image reconstruction means provides the reconstructed image as a frame signal to the switching block. 10. The method as claimed in claim 9, wherein the bandwidth limitation of the restored next frame is adaptively performed at a level of a predetermined plurality of levels corresponding to the degree of complexity of the recovered frame. system. 11. The method of claim 9 or 10, wherein the bandwidth limitation maps high frequency components below the determined bandwidth to zero for each transformed DCT transform coefficient block. ≪ RTI ID = 0.0 > system.