JPH11205792A - Block distortion reducing method and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To effectively suppress noise of block distortion, etc., even when a block size for encoding/decoding within a picture varies. SOLUTION: A noise reduction circuit 14 as this device is provided with a parameter arithmetic means 26 calculating a parameter required for deciding the block distortion of inputted picture data given encoding/decoding processing at each of plural kinds of blocks, a deciding means 30 judging block distortion based on a parameter from the means 26, a corrected value arithmetic means 25 calculating a corrected value showing a value for correcting picture data for reducing block distortion based on a judging result from the means 30 and a block distortion reducing means 34 reducing the block distortion of picture data based on the corrected value from the means 25.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a block for reducing block distortion that occurs when input data such as still image data or moving image data is subjected to block coding / decoding such as DCT coding. The present invention relates to a distortion reduction method and apparatus.

[0002]

2. Description of the Related Art Conventionally, block coding such as block DCT (discrete cosine transform) coding is known as a coding method for efficiently compressing and coding still image data, moving image data, and the like.

When encoding / decoding image data or the like of such an encoding method, ringing noise (mosquito noise) or block distortion may occur. As the compression rate increases, the input image increases. The more complicated, the easier it is to generate noise.

An apparatus for reducing mosquito noise and block distortion after encoding / decoding is based on macro characteristics at the block level and micro characteristics at the pixel level based on the local statistics of the image and the encoding information. A method of predicting the amount of noise and removing noise components by adaptive filtering is disclosed in Japanese Patent Application Laid-Open No. 7-236140. That is, here, in the case of the interlaced structure,
It discloses that by performing a process of removing a noise component on a field basis for an image having motion and on a frame basis for an image having little motion, more accurate noise reduction can be performed. I have.

[0005]

By the way, MPEG2
When encoding or decoding is performed by the scheme, when the picture structure is a frame structure, the DCT mode can be switched in macroblock units. That is, when performing encoding or decoding, it is possible to selectively switch between a frame mode in which encoding / decoding is performed for each frame in macroblock units and a field mode in which encoding / decoding is performed for each field. The DCT block for encoding or decoding is provided for each frame in the frame mode, and is provided for each field in the field mode.

That is, when encoding / decoding is performed in the MPEG2 system, encoding / decoding is performed in a macroblock unit, for example, in a field mode for a block having motion, and is performed in a frame mode for a block having no motion. I do.

[0007] When the size of a block to be coded / decoded in a screen is switched in this manner, block distortion is also generated according to the block. For this reason, if the block distortion reduction apparatus performs processing while fixing the size and position of the block, correction is performed up to a portion that is not the boundary of the block, and the block is broken or correction is performed on the boundary of the original block. There is a problem that it is not done.

Further, when different encoding / decoding is performed between the frame mode and the field mode, there is a problem that block distortion at a boundary between macroblocks cannot be effectively corrected.

Therefore, the present invention has been proposed in view of the above-mentioned situation, and even if the size of a block for encoding / decoding changes in a screen, block distortion and the like can be effectively reduced. It is an object of the present invention to provide a block distortion reduction method and apparatus capable of suppressing the noise of the block.

[0010]

A block distortion reducing method according to the present invention for solving the above-mentioned problems is required for determining block distortion of input image data which has been subjected to encoding / decoding processing for each of a plurality of types of blocks. Parameter calculating step of calculating various parameters according to the type of block, determining step of determining block distortion based on the result of the parameter calculating step, and correction value calculating step of calculating a correction value indicating a value for correcting image data And a block distortion reduction step of reducing block distortion of the image data based on the correction value.

In such a block distortion reduction method, the parameter calculation step is performed according to the type of input image data, so that the determination step, the correction value calculation step, and the block distortion reduction step are performed.

Further, the block distortion reduction apparatus according to the present invention calculates a parameter necessary for determining a block distortion of input image data which has been subjected to encoding / decoding processing for each of a plurality of types of blocks in accordance with the type of block. Parameter calculating means for performing the determination, a determining means for determining the block distortion based on the parameter from the parameter calculating means, and a correction value calculating means for calculating a correction value indicating a value for correcting the image data based on the determination result from the determining means And a block distortion reduction unit that reduces block distortion of image data based on the correction value from the correction value calculation unit.

This block distortion reduction device calculates a parameter according to the type of input image data,
Block distortion is reduced according to each type of image data.

[0014]

Embodiments of the present invention will be described below with reference to the drawings.

The recording system of the optical disk recording / reproducing apparatus 1 shown in FIG. National Television System Committe
e) Decoder 4 and Moving Picture Experts Group (MPEG) to which image data is input from NTSC decoder 4
p) Encoder 5, ECC (Error Correction Codes) encoder 6 to which image data is input from MPEG encoder 5, 8-14 modulation circuit 7 to which image data is input from ECC encoder 6, and 8-14 modulation circuit 7 And an RF amplifier 8 to which image data is input.

The A / D conversion circuit 3 performs an A / D conversion process on the NTSC image signal input from the input terminal 2. The A / D conversion circuit 3 converts an analog image signal into digital image data by performing an A / D conversion process. The A / D conversion circuit 3 outputs the image data to the NTSC decoder 4.

The NTSC decoder 4 receives NTSC image data from the A / D conversion circuit 3. The NTSC decoder 4 performs a decoding process so as to separate the input NTSC image data into a luminance signal and a chroma signal. The NTSC decoder 4 expands image data by performing a decoding process. Then, the NTSC decoder 4 outputs the image data to the MPEG encoder 5.

The MPEG encoder 5 adds the block DCT (Discrete Cos
ine Transform (Discrete Cosine Transform) Coding processing is performed. The MPEG encoder 5 converts the image data
The image data of the MPEG system is obtained by performing the CT encoding process. At this time, the MPEG encoder 5 adds encoding information such as a quantization scale to the image data to form a bit stream.

The MPEG encoder 5 has a pre-filter whose detailed configuration will be described later. This M
The PEG encoder 5 performs a filtering process using a pre-filter to reduce block distortion and mosquito noise generated in the compression-encoded image data. Then, the MPEG encoder 5 outputs the image data to the ECC encoder 6.

The ECC encoder 6 adds an error correction to the bit stream from the MPEG encoder 5. Then, the ECC encoder 6 outputs the bit stream to the 8-14 modulation circuit 7.

The 8-14 modulation circuit 7 subjects the bit stream from the ECC encoder 6 to signal processing such as 8-14 modulation. The 8-14 modulation circuit 7 outputs a bit stream subjected to 8-14 modulation or the like to the RF amplifier 8.

The RF amplifier 8 amplifies the bit stream from the 8-14 modulation circuit 7 and outputs it to the optical pickup 9.

The recording system of the optical disc recording / reproducing apparatus 1 records a bit stream indicating an image on the optical disc D via the optical pickup 9.

The reproducing system of the optical disk recording / reproducing apparatus 1 includes an RF amplifier 10 for inputting image data recorded on the optical disk D via an optical pickup 9 and an image amplifier for inputting image data from the RF amplifier 10. 14 demodulation circuit 1
1, an ECC decoder 12 to which image data is input from the 8-14 demodulation circuit 11, an MPEG decoder 13 to which image data is input from the ECC decoder 12,
A noise reduction circuit 14 to which image data is input from the decoder 13, an image quality correction circuit 15 to which image data in which noise is suppressed is input from the noise reduction circuit 14, and an image data subjected to image correction from the image quality correction circuit 15. NTSC encoder 16 to be input and NTSC encoder 16
And a D / A conversion circuit 17 to which NTSC image data is input.

The RF amplifier 10 amplifies the image data from the optical disk D detected by the optical pickup 9 and outputs the image data to the 8-14 demodulation circuit 11.

The 8-14 demodulation circuit 11 includes an RF amplifier 10
Is subjected to an 8-14 demodulation process. This 8-
The demodulation circuit 11 converts the demodulated image data into E
Output to the CC decoder 12.

The ECC decoder 12 performs an error correction process by using the error correction code added by the ECC encoder 6 by performing a decoding process on the image data from the 8-14 demodulation circuit 11. Then, the ECC decoder 12 outputs the image data subjected to the error correction processing to the MPEG decoder 13.

The MPEG decoder 13 decodes the MPEG image data from the ECC decoder 12. The MPEG decoder 13 outputs the decoded image data to the noise reduction circuit 14.

The noise reduction circuit 14 performs a noise reduction process on the image data from the MPEG decoder 13 by performing a filtering process, which will be described in detail later. The noise reduction circuit 14 performs noise reduction processing to reduce mosquito noise and block distortion caused by performing decoding processing by the MPEG decoder 13. Further, the noise reduction circuit 14 is connected to a control circuit 19 described later, and performs a control operation in accordance with a control signal from the control circuit 19. Then, the noise reduction circuit 14 outputs the image data subjected to the noise reduction processing to the image quality correction circuit 15.

The image quality correction circuit 15 includes a noise reduction circuit 14
Image quality correction processing is performed on the image data from. The image quality correction circuit 15 performs, for example, an outline correction process or the like as the image quality correction process. The image quality correction circuit 15 is connected to a control circuit 19 to be described later, and is controlled according to a control signal from the control circuit 19. The image quality correction circuit 15 converts the image data subjected to the image quality correction processing into NTS
Output to C encoder 16.

The NTSC encoder 16 performs an encoding process on the image data from the image quality correction circuit 15. This N
The TSC encoder 16 converts the image data into image data conforming to the NTSC system by performing an encoding process. The NTSC encoder 16 is adapted to use the NTS
The image data of the C system is output to the D / A conversion circuit 17.

The D / A conversion circuit 17 performs a D / A conversion process on the NTSC image data from the NTSC encoder 16. The D / A conversion circuit 17 performs a D / A conversion process to obtain an analog NTSC image signal. Then, the D / A conversion circuit 17 outputs the image signal subjected to the D / A conversion processing to the output terminal 18.

Further, the reproduction system of the optical disk recording / reproducing apparatus 1 has a control circuit 19 for supplying control signals to the above-described noise reduction circuit 14 and image quality correction circuit 15 and an input to the control circuit 19 operated by a user, for example. An operation input unit 20 for supplying a signal is provided.

The control circuit 19 comprises, for example, a microcomputer or the like, and supplies a control signal to the noise reduction circuit 14 or the image quality correction circuit 15 described above. This control circuit 1
9 supplies a control signal for reducing, for example, block distortion to the noise reduction circuit 14 according to an input signal from the operation input unit 20. The control circuit 19 supplies a control signal indicating whether or not to perform image quality correction and a control signal indicating the degree of image quality correction to the image quality correction circuit 15.

The operation input unit 20 generates and outputs an input signal by, for example, selectively pressing a switch or the like when a user or the like controls on / off of block distortion reduction. The operation input unit 20 is also provided with a switch or the like that allows the user to control the degree of image quality correction, and generates and outputs an input signal when the user selectively presses the switch.

Next, the configuration of the noise reduction circuit 14 will be described.

This noise reduction circuit 14 reduces the block distortion occurring at the block boundaries indicating the boundaries between the blocks, and as shown in FIG. 2, the HD / VD input terminal 2
1 and a control signal generating unit 22 for controlling each unit.
A delay circuit 24 that controls transmission timing of a signal from the chroma signal input terminal 23, a correction signal calculation unit 25 that generates a correction signal, and a parameter calculation unit 26 that calculates parameters necessary for determining block distortion. An edge extraction unit 27 for extracting an edge element of an image, a motion detection unit 28 for detecting a motion vector of the image, a vertical correlation detection unit 29 for detecting a correlation in the vertical direction, and a block for determining whether or not block distortion has occurred. And a distortion determination unit 30.

In the noise reduction circuit 14, the input terminals 23 and 31 are supplied with a video signal or a chroma signal and a luminance signal of image data which are decoded after image coding including block coding. . As a specific example of the image coding including the block coding, there is a so-called MPEG coding standard. This MPEG means ISO / IEC JTC1 / SC29 (Internationa
l Organization for Standardization / International
Electrotechnical Commission, Joint Technical Comm
ittee 1 / Sub Committee 29: Moving Picture Experts Gro, a study organization for video compression coding by the International Standards Organization / International Electrotechnical Commission Joint Technical Committee 1 / Special Committee 29)
up), ISO11172 as the MPEG1 standard,
ISO13818 is an MPEG2 standard. In these international standards, ISO11172-1
ISO13818-1 and ISO13818-1
18-2 is standardized, and ISO11172-3 and ISO13818-3 are standardized for audio.

Here, ISO1 as an image compression and coding standard is used.
In 1172-2 or ISO13818-2, an image signal is compression-encoded on a picture (frame or field) basis using the correlation in the temporal and spatial directions of the image.
The use of the spatial correlation is realized by using block DCT coding. In the following description,
Image data input to each of the input terminals 23 and 31 is input for each block to be encoded / decoded, and this block is referred to as a DCT block. Also, of the DCT blocks, DCs that are encoded / decoded in a frame structure are used.
A T block is called a frame DCT block, and a DCT block in which encoding / decoding is performed in a field structure is called a field DCT block.

As described above, after being subjected to compression coding including block DCT coding and recorded / reproduced, the chroma and luminance components of the image data inversely DCT-processed on the decoder side are converted to the chroma signal input terminal 23 and It is supplied to each of the luminance signal input terminals 31.

The luminance signal input to the luminance signal input terminal 31 is supplied to a correction signal calculating section 25 and a parameter calculating section 26.
Are sent to the edge extraction unit 27 and the motion detection unit 28.

The horizontal synchronizing signal and the vertical synchronizing signal are input from the HD / VD input terminal 21 to the control signal generator 22. The control signal generator 22 generates and outputs a timing signal necessary for driving each circuit based on the signal from the HD / VD input terminal 21.

The delay circuit 24 has a chroma signal input terminal 23
Performs delay processing on the chroma signal from. The delay circuit 24 outputs a chroma signal to a chroma signal output terminal 32 at a predetermined timing by performing a delay process.

The correction signal calculation unit 25 includes a correction value calculation circuit 3
3 and an adder 34. These correction value calculation circuits 33
The adder 34 receives a luminance signal from the luminance signal input terminal 31. Further, a timing signal is input from the control signal generator 22 to the correction value calculation circuit 33, and a product of weight coefficients Kc and Kp described later (Kcp = Kc × Kp)
Is entered.

The correction value calculation circuit 33 calculates the correction value by predicting the corrected inclination from the adjacent difference between the pixels on both sides of the boundary of each DCT block. The correction value calculation circuit 33
Then, a correction value is calculated based on the weighting coefficient from the block distortion determination unit 30, and a correction value for each pixel is calculated which is inversely proportional to the distance from the boundary. This correction value calculation circuit 3
At 3, the calculated correction value is output to the adder 34.

In the adder 34, the correction value from the correction value calculation circuit 33 and the luminance signal from the luminance signal input terminal 31 are added. The adder 34 performs a correction process on the input luminance signal by performing the addition process in this manner. The adder 34 outputs the corrected luminance signal to the luminance signal output terminal 44.

The parameter calculation section 26 has a parameter calculation circuit 35. The parameter operation circuit 35 receives a timing signal from the control signal generator 22 and a luminance signal from the luminance signal input terminal 31. The parameter calculation circuit 35 calculates an activity indicating an average value of absolute values of differences between adjacent pixels in each block described later. The luminance signal input to the parameter calculation circuit 35
0 is used to determine the necessary parameters in the correction block.

The edge extracting section 27 is connected to the luminance signal input terminal 3
A BPF (bandpass filter) 36 to which a luminance signal is input from 1; an absolute value circuit 37 to which a luminance signal is input from the BPF 36; a maximum value extraction circuit 38 to which a luminance signal is input from the absolute value circuit 37; And a binarization circuit 39.

The luminance signal input to the edge extraction unit 27 is input to the BPF 36, and a second differentiation is performed to detect an edge element of an image indicated by the input luminance signal. In the present embodiment, edge elements are extracted using, for example, Laplacian. And BPF
The luminance signal whose edge has been extracted at 36 is converted to an absolute value conversion circuit 37.
Is subjected to absolute value processing, and is input to the maximum value extracting circuit 38. The maximum value extraction circuit 38 detects the maximum value in an edge extraction block described later across a block boundary in order to obtain a threshold value required by the next-stage binarization circuit 39.

In the binarizing circuit 39, the maximum value extracting circuit 38
And the luminance signal extracted with the edge element by the BPF 36 are input, and the luminance signal is binarized based on the threshold. Then, the binarization circuit 39 outputs the binarized luminance signal to the vertical correlation detection unit 29.

The vertical correlation detecting section 29 includes an edge extracting section 27
The strength of the correlation in the vertical direction at the block boundary of the edge component extracted in (1) is obtained. That is, the vertical correlation detector 29 calculates the strength of vertical correlation based on the edge component from the edge extractor 27. Then, the vertical correlation detection unit 29 outputs the strength of the vertical correlation to the block distortion determination unit 30 as a weight coefficient Kc.

The motion detector 28 has a luminance signal input terminal 3
The luminance signal is input to the memory 40 from 1. In the motion detection unit 28, writing is performed on the memory 40 under the control of the memory controller 41. The luminance signal of the DCT block, which has been written before being read from the memory 40 by the memory controller 41, is input to the pattern matching unit 42, where the input luminance signal is subjected to pattern matching. The calculation result of the pattern matching is input to the motion vector determination unit 43, and the magnitude of the motion component is determined. Here, the presence or absence of the obtained motion component is input to the block distortion determination unit 30 as a motion coefficient.

The block distortion determining section 30 determines the block distortion based on the parameters in the correction block from the parameter calculating section 26, the strength of the vertical correlation from the vertical correlation detecting section 29, and the motion coefficient from the motion detecting section 28. Then, the weight coefficient of the block distortion correction value is calculated. The block distortion determination unit 30 outputs the weighting coefficient Kcp considering these factors to the correction value calculation circuit 33.

In the noise reduction circuit 14, an example has been described in which the processing for reducing the block distortion is performed only on the luminance signal. However, the same processing may be performed on the chroma signal.

Next, an example in which the noise reduction circuit 14 configured as described above performs processing for reducing block distortion is shown in FIG.
This will be described with reference to the flowchart of FIG. In the following description, an example in which the processing is performed in the H (horizontal) direction is shown. However, even in the processing in the V (vertical) direction, except that the following processing in the H direction is changed to the V direction. The description is omitted because it is the same.

In FIG. 3, first step ST1
In step 1, it is determined whether or not the block distortion reduction processing has been completed for all block boundaries in the H (horizontal) direction. If YES, the processing is terminated. If NO, the process proceeds to next step ST12. move on.

Here, the pixels used for the block distortion reduction processing will be described with reference to FIG. In the example of FIG. 4, for example, when performing block coding, D
This shows a specific example in which a frame DCT block is composed of 8 × 8 pixels using CT coding. That is, when four pixels are set as the correction range from the block boundary of the left and right DCT blocks 51L and 51R in FIG.
The block distortion correction processing block 53 is an eight pixel block, and is a block composed of eight pixels on one line centered on the block boundary. That is, in ST11 of the flowchart of FIG. 3, it is determined whether or not the block distortion correction processing has been performed for all the correction processing blocks 53.

In step ST12, the parameter operation circuit 35 determines the boundary difference | tmp0 | and the activity | tm as parameters required for determining whether or not there is block distortion.
p | and the adjacent difference | diff | are calculated by the following formulas.

| Tmp0 | = | fe | | tmp | = (| ba | + | cb | + | dc | + | ed | + | gf |
+ | Hg | + | ih | + | ji |) / 8 | diff2 | = | dc || diff3 | = | ed || diff4 | = | gf || diff5 | = | hg | Thus, the boundary difference | tmp0 |
Is the absolute value of the difference between the pixels e and f adjacent to each other across the boundary of the DCT block 51 in FIG.
tmp | is the average of the absolute values of the differences between adjacent pixels (except between e and f) for the block distortion processing block, and the adjacent difference | diff | is between pixels c and d, d and e.
, F, g, and the absolute value of each difference between g, h.

Next, in step ST13, the edge extraction unit 27 extracts edges in the edge processing extraction block 52 corresponding to the block distortion correction processing block 53. Then, in this step ST13, the vertical correlation detector 29 detects the strength of the vertical correlation. The details of the edge extraction processing will be described later.

Next, proceeding to step ST14, the motion detecting section 28 checks the motion coefficients of the two DCT blocks 51L and 51R sandwiching the block distortion correction processing block 53 to determine the presence or absence of motion. Details of the motion detection operation will be described later.

Next, in step ST15a, the block distortion judging section 30 uses the parameters obtained in steps ST12, ST13, and ST14, the strength of vertical correlation, and the motion coefficient to determine the block boundary. A determination is made as to whether or not there is block distortion and a process for determining the strength of correction is performed to calculate a correction value. An example of the block distortion determination process will be described later.

In the next step ST15b, if the block distortion determining section 30 determines that there is block distortion, the process proceeds to step ST16.
a, if it is determined that there is no block distortion, step ST18
Proceed to.

In step 16a, the block distortion determining section 3
0, the corrected boundary step | step | is obtained from the adjacent difference based on the property of the image, particularly the linearity, from the equation | step | = | diff3 + diff4 | / 2. Then, the correction amount | σ | necessary to have such a boundary step | step | after the correction is obtained by | σ | = (| tmp0 | − | step |) / 2. Here, it is preferable to switch the correction strength by discriminating the boundary difference | tmp0 | with a predetermined threshold corr_th. In this case, | tmp0 | <corr_th, and the correction strength is set to be strong (correction strength). When the correction amount | σ | is | σ | = (| tmp0 | − | step |) / 2 and | tmp0 | ≧ corr_th and the correction is weak, the correction amount | σ | , | Σ | = (| tmp0 | − | step |) / 4.

This is because if the boundary difference | tmp0 | is larger than the predetermined threshold value corr_th, there is a possibility that an erroneous judgment is made in the block distortion judgment even though an edge actually exists at the block boundary. In order to avoid this, the strength of the correction is switched between strong and weak.

In the next step ST16b, the correction value calculation circuit 33 obtains a correction value for each pixel from the obtained correction value | σ |. In order to smooth the joint with the adjacent correction range and because the block distortion appears more strongly near the boundary, a correction value inversely proportional to the distance from the boundary is obtained as shown in the following equation.

Specifically, when the correction values for the pixels _b to _{i in} the block distortion correction processing block 53 of FIG. 4 are | σ _b | to | σ _i |, respectively, the above correction value | σ | │σ _e │ = │σ│, │σ _f │ = │σ│ │σ _d │ = │σ│ / 2, │σ _g │ = │σ│ / 2 │σ _c │ = │σ│ _{/ 4, | σ h | =} | σ | / 4 | σ b | = | σ | / 8, | σ i | = | Request / 8 of each such correction values respectively | sigma.

In the next step ST16c, the correction value | σ for each of the pixels b to i obtained in the above step ST16b
_The video signals (image data) SB _{b to} SB _i that have been subjected to the block distortion correction are obtained using _b | to | σ _i |.

More specifically, when the input image data of each of the pixels _{b to i} before correction is S _{b to} S _i , the corrected image data SB _{b to} SB _i are calculated according to the sign of tmp0. _{tmp0 ≧ 0: SB b = S} b + | σ b |, tmp0 <0: SB b =
_{S b - | σ b | tmp0} ≧ 0: SB c = S c + | σ c |, tmp0 <0: SB c =
_{S c - | σ c | tmp0} ≧ 0: SB d = S d + | σ d |, tmp0 <0: SB d =
_{S d - | σ d | tmp0} ≧ 0: SB e = S e + | σ e |, tmp0 <0: SB e =
S _e − | σ _e | tmp0 ≧ 0: SB _f = S _f − | σ _f |, tmp0 <0: SB _f =
_{S f + | σ f | tmp0} ≧ 0: SB g = S g - | σ g |, tmp0 <0: SB g =
_{S g + | σ g | tmp0} ≧ 0: SB h = S h - | σ h |, tmp0 <0: SB h =
S _h + | σ _h | tmp0 ≧ 0: SB _i = S _i − | σ _i |, tmp0 <0: SB _i =
A correction value calculating circuit 33 and an adder 34 perform correction such that S _i + | σ _i |
Do with.

In the next step ST17, the signal subjected to the block distortion correction processing is output from the luminance signal output terminal 44.

If it is determined in step ST15b that there is no block distortion, the process proceeds to step ST18, and the original signal in the correction range is output as it is.

Next, an example of the edge extraction processing in step ST13 will be described with reference to FIG.

When performing the edge extraction processing, in step ST21, the BPF 36 of the edge extraction unit 27 and the absolute value conversion circuit 37 take the absolute value of the secondary differential signal, and the process proceeds to step ST22.

Next, in step ST22, the maximum value is detected in the processing block by the maximum value extracting circuit 38, and the process proceeds to step ST23.

Next, using this maximum value, step ST2
In 3, the binarization circuit 39 binarizes the BPF-processed image and performs edge detection. The threshold for binarization by the binarization circuit 39 may be, for example, 例 え ば of the maximum value obtained by performing the secondary differentiation and the absolute value processing in the block.

That is, in the edge extraction unit 27 of FIG.
The luminance signal from the luminance signal input terminal 31 is secondarily differentiated by the BPF 36 as described above, the absolute value is obtained by the absolute value conversion circuit 38, and the maximum value is detected by the maximum value detection circuit 35. Then, the threshold value from the maximum value detection circuit 35 is converted to a binarization circuit 36.
And the signal from the absolute value conversion circuit 38 is binarized. An output from the binarization circuit 36 is output to a vertical correlation detection unit 39.
Sent to

The vertical correlation detecting section 39 obtains the strength of the vertical correlation at the block boundary of the edge component thus extracted. An example of a method for obtaining the strength of the vertical correlation at this time will be described with reference to FIG.

In FIG. 6, a region b including a target block boundary and regions a and c near the region b are divided.
The number of edge components extracted previously is calculated for each region. These are Ea, Eb, and Ec. In FIG. 6, a pixel extracted as an edge is denoted by 1 and a pixel determined to be not an edge is denoted by 0. In this example, Ea = 5,
Eb = 12 and Ec = 5.

The ratio Kv of the number of edge components in the region including the block boundary and the region not including the block boundary is obtained, and the classification is performed (step ST24 in FIG. 5).

For example, when Kv ≧ 4, class 12 when 2 ≦ Kv <4, class 21 when 1 ≦ Kv <2, class 3 when Kv <1, and class 4. In the example of FIG. 6, Kv = (2 × 12) / (5 + 5) = 2.4, so that the class 2 is set.

Next, the block distortion determining section 30 shown in FIG. 2 will be described. In the block distortion determination section 30, a weighting coefficient Kc is set according to each class detected by the vertical correlation detection section 29.
Assign. The weight coefficient Kc for each class is, for example, a class weight coefficient Kc 1 1 2 0.75 3 0.5 4 0.25.

On the other hand, correction OFF / weak / strong judgment is made from each parameter value obtained by the parameter calculation unit 26, and
The weight coefficient Kp is obtained.

Correction stage Weight coefficient Kp OFF 0 Weak 0.5 Strong 1 The product of the weight coefficients Kc and Kp (Kcp = Kc × Kp) is sent to the correction value calculation unit 33 to control the block distortion correction amount.

Therefore, when the vertical correlation at the block boundary is strong, the correction amount increases, and the block distortion can be effectively removed. That is, the detection accuracy of the block distortion is improved.

Further, the class obtained from the vertical correlation detecting section 29 is used to determine the division threshold div_
th may be provided to control the value of the threshold div_th. For example, as the vertical correlation is weaker, the possibility of block distortion is lower. Therefore, the value of the detection threshold div_th is increased, and control is performed in a direction in which detection is difficult.

In the above example, the class and the correction step are described as four and three steps, respectively. However, the present invention is not particularly limited to this. For example, the weight coefficient Kc may be obtained by Kc = Eb / (Ea + Eb + Ec). .

The edge extracting unit 27, the correction signal calculating unit 25, and the block distortion determining unit 3 according to the embodiment of the present invention.
The algorithm described above was used for the parameter calculation unit 26 and the parameter calculation unit 26. However, the algorithm is not limited to these algorithms.
Various modifications such as using an LPF for 5, various edge extraction methods such as an edge extraction method by edge tracking for the edge extraction unit 27, and various parameters for the block distortion determination unit 30 can be considered.

The above embodiment is an example in which the block distortion correction is performed in the horizontal direction of the luminance signal. However, the present invention is not limited to this example. Various modifications such as applying a correction are possible.

As is clear from the above description, according to the embodiment of the block distortion reduction method and apparatus according to the present invention, while performing block distortion reduction without loss of high-frequency components and maintaining resolution, correction by block correction is performed. Failure can be reduced. In addition, since the hardware configuration is simple, not only for business use, but also various consumer devices that perform compression processing using block coding such as DCT coding, for example, video CD players,
It can be mounted on digital video disc players, digital television receivers, video phones, and the like. Needless to say, the above-described algorithm can be realized by software processing, and block distortion reduction and block distortion removal in so-called real-time reproduction of moving images on the Internet or multimedia can be easily realized. Further, according to the present embodiment, three modes of strong / medium / weak are provided, so that block distortion can be reduced according to the state of the video. Further, since the parameters used in the block distortion processing can be adjusted from the outside, fine adjustment is possible in addition to the above three modes.

Next, in the noise reduction circuit 14 described above,
A processing method in the case where the size of each block of the block encoding differs within a frame will be described.

Here, in the case of encoding according to the MPEG2 system, when the picture structure is a frame structure,
The CT mode can be switched between a frame mode and a field mode on a macroblock basis. 7 and 8 show the DCT mode (frame D) for the luminance signal.
(CT / field DCT). 7 and 8, the field DCT block is the frame DC
It has twice the spatial size of the T block in the vertical direction.

For example, as shown in FIG.
The case where the noise reduction processing is performed by the noise reduction circuit 14 when the T block and the frame DCT block are adjacent to each other will be described.

In general, a field DCT block is
Used when there is movement in the frame. That is, this is a case where the image in the left macroblock in FIG. 9 is stationary and the image in the right macroblock is mainly moving.
Thus, the field DCT block and the frame DCT
If the block is horizontally adjacent,
As shown in FIG. 9, the vertical correlation detection and the correction of the block distortion are performed by making the correlation detection area in the frame DCT block correspond to the correlation detection area of the adjacent field DCT block.
0 and the correction signal calculation unit 25.

When the field DCT block and the frame DCT block are adjacent to each other in the horizontal direction, a correlation detection area Ea in the frame DCT block is set as shown in FIG. 10, and Kc = 2Eb / ( Ea + 2Eb + 2Ec) Accordingly, the weight coefficient Kc is calculated. A weighting coefficient may be calculated according to such an expression, and the correction signal calculation unit 25 may perform correction processing on the luminance signal.

Therefore, by detecting the correlation and correcting the luminance signal in this way, even if there is a large difference in the amount of block distortion between the fields, the vertical correlation obtained by using the vertical correlation detection area shown in FIG. There is no possibility that the correlation between the correlation value and the actual magnitude of the block distortion becomes low and the effective block distortion reduction processing cannot be performed. FIG. 11 shows an example of a conventional correlation detection method, and shows an example in which a frame DCT block and a field DCT block are adjacent in the horizontal direction H.

Next, an example in which the noise reduction circuit 14 performs noise reduction processing when the field DCT block and the frame DCT block are adjacent to each other in the vertical direction will be described.

Field DCT block and frame DC
When the T block is vertically adjacent, FIG.
As shown in FIG. 2, the correlation detection area in the frame DCT block is made to correspond to the correlation detection area of the adjacent field DCT block, and the correlation detection in the unit of the field DCT block is performed by the vertical correlation detection unit 29 to correct the block distortion. The processing is performed by the block distortion determination unit 30 and the correction signal calculation unit 25.

By performing the block distortion reduction processing as described above, for example, when there is a large difference in the amount of block distortion between fields, FIGS. 13A and 13B
As shown in (1), there is no possibility that the correlation between the horizontal correlation value obtained using the horizontal correlation detection area and the actual magnitude of the block distortion becomes low, so that effective block distortion reduction processing cannot be performed. Further, by performing the block distortion reduction processing as described above, when the horizontal correlation detection area as shown in FIG. 14 is used, the edge extraction between the fields of the field DCT block is performed in the vertical direction, and the motion is reduced. If there is, accurate edge extraction cannot be performed.

By the way, in the edge extraction in the horizontal correlation detection processing, for example, in the frame DCT block, the transfer function V (z) of the secondary differential characteristic is calculated as follows: V (z) = (− 1 + 2z− ^H− z− ^2H ) / 4 In the field DCT block, the transfer function V (z) of the secondary differential characteristic is set as V (z) = (− 1 + 2z− ^2H− z− ^4H ) / 4.

That is, in the present embodiment, when the frame DCT block and the field DCT block are adjacent to each other, the correlation detection area when performing the correlation detection processing by the vertical correlation detection unit 29 is set to the value of the adjacent field DCT block. In addition to the correlation with the correlation detection area, the edge extraction processing by the edge extraction unit 27 is also performed in correspondence with the field DCT block, and is performed for each field.

Note that the size of the above-mentioned correlation detection area is
The present invention is not limited to the above example, and the areas may overlap each other, for example.

Next, another method of calculating the activity, which is calculated by the parameter calculator 26, will be described.

In another calculation method of this activity, the above calculation method calculates on the same line. However, as shown in FIG. 15, an activity calculation area is provided in the frame DCT block to provide a two-dimensional space. Calculate the activity. In such an activity calculation area, the activity | tmp | is calculated according to the following equation 1.

[0104]

(Equation 1)

Here, the block boundary of interest is the pixel e
This is the boundary between (0) and f (0). Then, the activity calculation area to be calculated also moves according to the block boundary.

In Expression 1, w [m] [n] indicates a weight coefficient, and | diff [m] [n] | indicates an adjacent difference. In Equation 1, the sizes M and N of the two-dimensional parameter space when calculating the activity are set to 7 and 8, respectively. This parameter is a difference between pixels around the block boundary, and the parameter space is not the activity calculation area itself.

This weight coefficient is expressed as shown in the following equation (2).

[0108]

(Equation 2)

The adjacent difference | diff | is expressed by the following equation (3).

[0110]

(Equation 3)

That is, by applying the weighting coefficients and the adjacent differences expressed by the equations (2) and (3) to the above equation (1),
The activity | tmp | can be calculated in the two-dimensional space. Therefore, in the parameter calculation unit 26, by expanding the activity calculation area to the two-dimensional space in this way, the block distortion determination unit 30 can more accurately determine the block distortion.

On the other hand, the activity | tmp | for the field DCT block is calculated by the following equation 4 by setting an activity calculation area as shown in FIG.

[0113]

(Equation 4)

Here, fi [m] [n] in Expression 4 is an arithmetic expression for corresponding to a field DCT block as shown in Expression 5 below.

[0115]

(Equation 5)

Then, in this equation (4), the above equation (5) is used.
The activity is calculated by using fi [m] [n] by using pixels for each field. Thereby, the parameter calculation unit 26
Then, for example, if the image has motion, the field DCT
When coding is performed using blocks, even when there is no correlation between fields, the block distortion determination unit 30 can more accurately determine block distortion.

When the parameter calculator 26 calculates an activity for vertical block distortion in a frame DCT block, the parameter calculator 26 sets an activity calculation area as shown in FIG. Weighting factor w [m] expressed by Equation 2 and Equation 3
[n], applying the adjacent difference | diff | in the vertical direction,
Equation 1 can be applied to calculate the vertical activity | tmp |.

When calculating the vertical activity in the field DCT block, the parameter calculation unit 26 sets an activity calculation area as shown in FIG. The activity | tmp | is calculated using fi [M] [N] of the following equation 6 instead of fi [M] [N].

[0119]

(Equation 6)

The parameter calculation unit 26 calculates fi
By applying [M] [N] to Equation 4, the actual activity calculation area is narrowed. This is because in the field DCT block, the distance between pixels in the vertical direction is large, so that spatial balance is achieved. That is, the parameter calculation unit 26 determines that the activity calculation area in the two-dimensional space is substantially the same in the frame DCT block and the field DCT block,
tmp | is calculated.

Next, an example of calculating the activity by the parameter calculation unit 26 when the frame DCT block and the field DCT block are adjacent in the vertical or horizontal direction will be described.

FIG. 19 is a diagram showing a state where a frame DCT block and a field DCT block are adjacent in the horizontal direction, and FIG. 20 is a diagram showing a state where they are adjacent in the vertical direction. When the parameter calculating unit 26 calculates an activity for an image as shown in FIG. 19, the activity is calculated by applying the above-described activity calculation method of the frame DCT block to the activity calculation method of the field DCT block. That is, in this case, the parameter calculation unit 26
Calculates the activity using the above equation 1.

The method of calculating the activity calculated by the parameter calculating unit 26 is not limited to the above-described example. The size of the activity calculation area, the weight coefficient of the difference between pixels around the block boundary in the two-dimensional space, Can be arbitrarily changed.

Further, in the above-described embodiment, the activity value is used as the criterion for determining the block distortion in the block distortion determination section 30. If the target block boundary step is smaller than the activity value, it is determined that the block distortion is not the block distortion. However, the block distortion was not corrected.
As shown in (1), the weighting coefficient K _act may be determined according to the ratio between the activity value and the block boundary difference, and the correction value of the block boundary step may be obtained. That is, as shown in FIG. 21, the parameter calculation unit 26 can prevent the correction near the threshold or the non-correction by continuously changing the value of the activity.

[0125]

As described above in detail, the block distortion reduction method according to the present invention provides a parameter necessary for judging the block distortion of input image data which has been subjected to encoding / decoding processing for each of a plurality of types of blocks. A parameter calculation step of calculating according to the type of the block, a determination step of determining block distortion based on a result of the parameter calculation step, and a correction value indicating a value for correcting image data to reduce the block distortion. Correction value calculation step and a step of reducing block distortion of image data based on the correction value, so that, for example, even if blocks of different sizes are input adjacently, the parameters are calculated according to the type of each block. Then, the block distortion can be determined to reduce the block distortion.

Further, the block distortion reducing apparatus according to the present invention sets parameters necessary for judging block distortion of input image data subjected to encoding / decoding processing for each of a plurality of types of blocks in accordance with the types of the blocks. Since parameter calculation means for calculating is provided, it is possible to reduce the block distortion of image data by calculating parameters necessary for the block distortion determination according to the input block and determining the block distortion by the determination unit. it can.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an optical disk recording / reproducing apparatus.

FIG. 2 is a block diagram illustrating an example of a noise reduction circuit to which the present invention has been applied.

FIG. 3 is a flowchart illustrating an example in which a noise reduction circuit to which the present invention is applied performs processing for reducing block distortion.

FIG. 4 is a diagram for describing pixels used for block distortion reduction processing.

FIG. 5 is a flowchart illustrating an example of an edge extraction process.

FIG. 6 is a diagram illustrating an example of a method for obtaining the strength of vertical correlation.

FIG. 7 is a diagram for explaining a frame DCT block in a luminance signal.

FIG. 8 is a diagram for explaining a field DCT block in a luminance signal.

FIG. 9 shows a field DCT block and a frame DCT.
FIG. 9 is a diagram for explaining an example of a case where noise reduction processing is performed by a noise reduction circuit 14 when blocks are adjacent to each other.

FIG. 10 shows a field DCT block and a frame DCT.
FIG. 9 is a diagram for explaining an example of setting a correlation detection area Ea in a frame DCT block and calculating a weight coefficient when a block is adjacent in the horizontal direction.

FIG. 11 is a diagram for explaining a vertical correlation detection area in a conventional method.

FIG. 12 shows a field DCT block and a frame DCT.
FIG. 11 is a diagram for explaining an example of calculating a weight coefficient when a block is adjacent to a block in the vertical direction.

FIG. 13 shows a field DCT block and a frame DCT.
FIG. 11 is a diagram for describing an example of calculating a weight coefficient by a conventional method when a block is adjacent to a block in the vertical direction.

FIG. 14 is a diagram for explaining an example when edge extraction between fields of a field DCT block is performed in the vertical direction when a horizontal correlation detection area is used.

FIG. 15 is a diagram for explaining that an activity calculation area is provided in a frame DCT block to calculate an activity in a two-dimensional space.

FIG. 16 is a diagram for explaining that an activity calculation area is provided in a field DCT block to calculate an activity in a two-dimensional space.

FIG. 17 is a diagram for explaining an example of calculating an activity for vertical block distortion in a frame DCT block.

FIG. 18 is a diagram for describing an example of calculating a vertical activity in a field DCT block.

FIG. 19 shows a frame DCT block and a field DCT.
It is a figure showing signs that a block is adjacent in the horizontal direction.

FIG. 20 shows a frame DCT block and a field DCT.
FIG. 3 is a diagram illustrating a state in which blocks are vertically adjacent to each other.

FIG. 21 is a diagram illustrating a weighting coefficient K _act that changes according to a ratio between an activity value and a block boundary difference.

[Description of Signs] 1 optical disk recording / reproducing device, 5 MPEG encoder, 13 MPEG decoder, 14 noise reduction circuit,
25 correction signal calculator, 26 parameter calculator, 27
Edge extractor, 28 motion detector, 29 vertical correlation detector, 30 block distortion determiner

Claims (16)

[Claims] A parameter calculation step of calculating parameters required for determining block distortion of input image data which has been subjected to encoding / decoding processing for each of a plurality of types of blocks in accordance with the type of the block; A determining step of determining the presence or absence of block distortion based on a result of the step; a correction value calculating step of calculating a correction value indicating a value for correcting the image data; and reducing a block distortion of the image data based on the correction value. A block distortion reduction method, comprising: a block distortion reduction step. 2. An edge extraction step of generating edge information by performing an edge extraction process of extracting an edge of the image data; and a motion detection process of performing a motion detection process of detecting a motion component of the image data. 2. The block distortion reduction according to claim 1, further comprising: determining whether or not there is block distortion based on a result of the edge information, the motion information, and the parameter calculation step. Method. 3. The method according to claim 1, wherein the plurality of types of blocks have at least different sizes. 4. The method according to claim 1, wherein the plurality of types of blocks are at least a frame coding / decoding block and a field coding / decoding block. 5. In the parameter calculation step, when a frame coding / decoding block and a field coding / decoding block perform a parameter calculation of adjacent image data, the parameter calculation corresponding to the field coding / decoding block is performed by the frame. The block distortion reduction method according to claim 4, wherein the parameter calculation is performed by applying to the encoding / decoding block. 6. The method according to claim 1, wherein the parameter calculating step calculates an activity indicating an average value of absolute values of differences between adjacent pixels in each block as a parameter necessary for determining block distortion. 3. The method for reducing block distortion according to item 1. 7. The block distortion reduction method according to claim 1, wherein the parameter computation step computes a spatial correlation of block distortion. 8. The correction value calculating step according to claim 6, wherein the correction value is calculated according to a ratio between the activity calculated in the parameter calculating step and a boundary step of each block. The method for reducing block distortion according to the above. 9. Parameter calculating means for calculating parameters necessary for determining block distortion of input image data subjected to encoding / decoding processing for each of a plurality of types of blocks in accordance with the type of the block, and the parameter calculation Determining means for determining the presence or absence of block distortion based on a parameter from the means; correction value calculating means for calculating a correction value indicating a value for correcting image data based on a determination result from the determining means; A block distortion reduction device comprising: a block distortion reduction unit configured to reduce block distortion of image data based on a correction value from a calculation unit. 10. An edge extracting means for generating edge information by performing an edge extracting process on the image data, and a motion detecting means for generating motion information by performing a motion detecting process for detecting a motion component of the image data. 10. The apparatus according to claim 9, wherein the determining unit determines whether there is block distortion based on the edge information, the motion information, and the parameter. 11. The parameter calculating means calculates a parameter necessary for determining a block distortion of input image data which has been subjected to encoding / decoding processing for at least a plurality of types of blocks having different sizes, the size of each block. The block distortion reduction device according to claim 9, wherein the calculation is performed according to: 12. The method according to claim 12, wherein the parameter calculation means includes at least a frame coding / decoding block and a field coding /
10. The block distortion according to claim 9, wherein parameters necessary for determining block distortion of input image data subjected to encoding / decoding processing for each decoding block are calculated according to the size of each block. Reduction device. 13. The parameter calculating means, when calculating parameters of image data adjacent to a frame encoding / decoding block and a field encoding / decoding block, performs a parameter operation according to the field encoding / decoding block on the frame. The block distortion reduction apparatus according to claim 12, wherein the parameter calculation is performed by applying the parameter calculation to an encoding / decoding block. 14. The apparatus according to claim 9, wherein said parameter calculation means calculates an activity indicating an average value of absolute values of differences between adjacent pixels in each block as a parameter necessary for determining block distortion. 3. The block distortion reduction device according to item 1. 15. The apparatus according to claim 9, wherein said parameter calculating means calculates a spatial correlation of block distortion.
3. The block distortion reduction device according to item 1. 16. The correction value calculation means according to claim 14, wherein said correction value is calculated in accordance with a ratio between said activity calculated by said parameter calculation means and a boundary step of each block. The block distortion reduction device as described in the above.