JPH02192284A - Quantization circuit for picture data - Google Patents

Abstract

PURPOSE:To prevent visual deterioration in a picture quality of a decoded picture by applying quantization so that a decoded picture signal has a quantity change closest to a spatial change quantity of an original picture data. CONSTITUTION:A difference between a true value x1 of a noted picture element and a true value x0 of a surrounding picture element such as a preceding picture element (change in horizontal direction) DELTAr(=x1-x0) is detected. A decoding level X0 of the picture element x0 is generated by a local decoder 52 and 4 decoding representative levels (I0+MIN, I1+MIN, I2+MIN, I3+MIN) are generated by local decoders 48-51. Then subtraction circuits 40-43 calculate a difference between the decoding representative levels and the decoding level X0 (predicted change) DELTA0, DELTA1, DELTA2, DELTA3 are calculated. Moreover, a change quantity closest to a change DELTAr of the original picture data is detected among the 4 differences and a quantization code in response to the detection is selected by a code selection circuit 39. Thus, the deterioration distinct visually is prevented.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image data quantization circuit that is applied to compress and encode image data.

[Summary of the invention]

In this invention, in an image data quantization circuit that compresses digital data quantized with n bits into data with m bits (n>m), codes of upper m bits of digital data of n bits are spatially adjacent to each other. or between two sample data that correspond spatially and are consecutive in time, and detect that both of these two sample data take a level near the change point of the m-bit code. A detection circuit, the amount of spatial change between the true value of the pixel of interest and the true value of reference pixels surrounding the pixel of interest, or the true value of the pixel of interest, and the amount of change in space between the true value of the pixel of interest and the true value of reference pixels surrounding the pixel of interest, and Detecting the amount of temporal change between the true value of the spatially corresponding reference pixel, and detecting the amount of spatial or temporal change of the reference pixel with respect to the decoded value of the quantization code. A quantization circuit that calculates a decoded representative value closest to the amount of change and selects a quantization code corresponding to the decoded representative value, m-bit digital data formed from n-bit digital data, and a quantum By providing a selection circuit that selectively outputs the output of the conversion circuit and the output signal of the detection circuit according to the output signal of the detection circuit, visually noticeable noise is prevented from occurring in the restored image, and the level corresponds well with the original data. output data is obtained.

[Prior Art] Various encoding methods have been proposed in which the number of bits of each pixel (sample) of digitized image data is reduced by using the correlation of image signals. The applicant of this application is
As described in the specification of No. 266407, a high-efficiency encoding method calculates a dynamic range, which is the difference between the maximum value and minimum value of multiple pixels included in a two-dimensional block, and performs encoding adapted to this dynamic range. We are proposing a device. In addition, as described in Japanese Patent Application No. 60-232789, high-efficiency encoding is performed that performs encoding adapted to the dynamic range of a three-dimensional block formed from pixels of multiple regions each belonging to multiple frames. A device has been proposed.

Furthermore, as described in Japanese Patent Application No. 60-268817, there is a variable length encoding method in which the number of bits changes depending on the dynamic range so that the maximum distortion caused when quantization is constant. is proposed.

Coding adapted to these dynamic ranges (hereinafter referred to as
(abbreviated as ADRC) is a high-efficiency encoding method that reduces the number of bits per pixel by utilizing the fact that images have strong correlation in small areas (blocks) into which one screen is divided. In other words, the difference between the minimum or maximum value in the block and the level of each pixel is smaller than the original level, and this difference is reduced to a number of bits (e.g. 8 bits) smaller than the original number of bits (e.g. 8 bits).
For example, 4 bits) can be used for quantization.

This invention can be applied to the above-mentioned quantization of the level normalized by the maximum value or minimum value in ADRC, or to compressing an n-bit code obtained by ADRC to m bits (n > m). .

However, this invention is not limited to ADRC,
The present invention can be applied to a quantization circuit that compresses an n-bit digital image signal into m-bits.

- For example, in ADRC that performs 2-bit quantization, the dynamic range DR of the block, which is the difference between the maximum value MAX and the minimum value MIN, is divided equally into four level ranges, and after the minimum value MIN is removed, The pixel values of are expressed by 2-bit quantization codes corresponding to the level ranges. On the decoding side, one of the central decoding representative levels 10-13 of each level range is decoded from the dynamic range DR and the quantization code, and the minimum value MIN is added to the decoded value to determine the pixel data in the block. is restored.

FIG. 6 shows an example of quantization in ADRC.
The figure shows an example of one-dimensional ADRC in which one block consists of six pixels consecutive in the horizontal direction. The data shown by white dots is the true value of 4 bits, 8 bits, etc. of the pixels in the block, and the solid line shows the real value of the pixels in the block. It has a horizontal change indicated by 101. In the case of encoding with the above-mentioned 2-bit ADRC, a restoration level indicated by x is obtained on the decoding side, and a change in the signal indicated by a broken line 102 occurs in the restored image.

FIG. 7 shows another example of quantization in ADRC. FIG. 7 belongs to each of six consecutive frames in the time direction,
In addition, since 0wi indicates the time change of pixels at spatially corresponding positions, it is assumed that each block including these six pixels has the same maximum value MAX and minimum value MEN. . The data indicated by white dots is the true value of the pixel, and has a change in the time direction indicated by a solid line 103. In the case of encoding with the above-mentioned 2-bit ADRC, a restoration level indicated by × is obtained on the decoding side,
In the restored image, a signal change indicated by a broken line 104 occurs.

[Problem B to be Solved by the Invention] In conventional quantization, the original pixel level is replaced with the closest decoded representative level in order to reduce the quantization error and improve the S/N ratio. However, even if it is quantitatively good, visually noticeable deterioration may occur in the restored image. In the example shown in FIG. 6, the original gentle horizontal change 101 becomes a severe change 102 after restoration, and visually noticeable noise (deterioration) occurs in the restored image. This noise is like a finer version of the snow noise that occurs in images received by television broadcasts during weak electric fields, and is a shrill noise. This problem occurs because humans are sensitive to differential characteristics of images when recognizing them.

Similar to the spatial change, in the example shown in FIG. 7, the original gradual change 103 in the time direction becomes a severe change 104 after restoration, and the same noise as described above occurs in the restored image.

As can be seen from FIGS. 6 and 7, the above-mentioned noise occurs when the data of two spatially adjacent or spatially corresponding but temporally consecutive samples of the original digital data are different, and one of them is different from the other. This is due to the fact that one bit is quantized into a 2-bit code different from the other. In other words, these two samples
- Noise occurs when the noise is included above and below the boundary of two adjacent level regions, respectively.

Therefore, an object of the present invention is to provide an image data quantization circuit that can preserve the spatial variation of the original image signal and improve the visual quality of the restored image even if quantitative errors increase. Our goal is to provide the following.

Another object of the present invention is to quantize image data so that output data that closely corresponds in level to the original data can be obtained by performing quantization that takes visual characteristics into consideration only when there is a risk of noise occurring. The purpose is to provide circuits.

[Means to solve the problem]

In this invention, in an image data quantization circuit that compresses digital data quantized with n bits into data with m bits (n>m), codes of upper m bits of digital data of n bits are spatially adjacent to each other. Detection that detects that there is a difference between two sample data, or two spatially corresponding and temporally continuous sample data, and that both of these two sample data take a level near the changing point of the m-bit code. The circuit 19 and the spatial variation amount Δ between the true value of the pixel of interest and the true values of reference pixels surrounding the pixel of interest.
r, or the amount of temporal change between the true value of the pixel of interest and the true value of a reference pixel that temporally precedes and spatially corresponds to the pixel of interest, and decodes the quantization code. Find the decoded representative value where the spatial or temporal change amount Δ0 to Δ3 of the reference pixel with respect to the value is closest to the detected spatial or temporal change amount Δr, and select the quantization code corresponding to the decoded representative value. A circuit 13 that selectively outputs m-bit digital data formed from n-bit digital data and the output of the quantization circuit 21 according to the output signal of the detection circuit 19. are provided.

[Effect]

4-bit quantization code D generated during ADRC encoding
In order to further compress T, the upper two bits of the quantization code DT are selected. These upper 2 bits and 2 bits from the quantization circuit 21 that takes visual characteristics into account are used in the switch circuit 1.
3 is selected. The switch circuit 13 is controlled by the output signal of the detection circuit 19. The detection circuit 19 detects that two spatially adjacent sample data or two spatially corresponding and temporally continuous sample data are different and included in different level regions in the case of 2-bit quantization. . At the time of this detection, the output signal of the quantization circuit 21 is selectively output.

A case will be described in which the quantization circuit 21 forms output data having the same spatial variation as the original data. The decoding level xO of the 0 pixel XO, where the difference (horizontal change amount) Δr (-xi-xO) between the true value x1 of the pixel of interest and the true value XO of a surrounding pixel, for example, the previous pixel, is detected is generated by the local decoder. be done. In addition, 4 decoding representative levels (10+
MIN.

11+MIN, 12+MIN, 13+MIN) are generated by the local decoder. In the subtraction circuits 40 to 43, the difference between the decoding representative level and the decoding level XO (predicted change 1
1) ΔO1Δ1, Δ2, and Δ3 are calculated. That is, Ao-(IO+MIN)-XO AI-(11+MIN)-XO Δ2- (12+MIN)-XO Δ3- (130 MIN)-XO Among these four differences, the original image data has The amount of change that is closest to the amount of change Δr that was previously detected is detected. Therefore, in the subtraction circuits 34 to 37, (β0−Δr−ΔO
)(β1−Δr−Δ1)(β2−Δr−Δ2)(β3−
A subtraction process of Δr−Δ3) is performed, the minimum subtraction output is detected, and a quantization code corresponding to this detection is selected by the code selection circuit 39. That is, when the difference with ΔO is the smallest, the quantization code of (00) is selected, and thereafter the quantization codes of (01), (10), and (11) are selected in the same manner.

As mentioned above, quantization is performed so that the restored image signal has the spatial variation that is closest to the spatial variation of the original image data, thereby preventing visual deterioration of the image quality of the restored image. Can be done. Further, when there is no risk of visually noticeable noise occurring in the restored image, quantitative quantization is performed, so output data that is faithful in level to the original data can be obtained.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings. This description is given in the following order.

a. Overall configuration and operation of one embodiment, quantization circuit C1 modification example a. Overall configuration and operation of one embodiment In FIG. A digital video signal in which each pixel (one sample) is quantized to 8 bits is supplied. The blocking circuit 2 converts the data order of the input digital video signal from the scanning order to the block order. For example, one frame of screen is subdivided into (4X4-166 pixels) Blocks are constructed. In Fig. 2, N-1 indicates the previous block, and N indicates the block of interest to be encoded. The data for the top pixel is transmitted first, then the data for the three pixels lined up in the horizontal direction are transmitted, the data is transmitted in the same way for the second line, and finally the bottom pixel on the right side is transmitted. pixel data is transmitted.

The output signal of the blocking circuit 2 is supplied to the maximum value and minimum value detection circuit 3, and the maximum skin grafting MAX and minimum value MIN of the pixels included in each block are respectively detected. Maximum value MA
X and the minimum value MIN are supplied to a subtraction circuit 6, and a dynamic range DR, which is the difference between the two, is calculated. Dynamic range DR and minimum value MIN are each delay circuit 8.9
It is taken out to the output terminal 10.11 via. A framing circuit (not shown) is connected to the output terminals 10, 11, and 16, and the framing circuit has a dynamic range D.
R, the minimum value MIN, and a code signal compressed to 2 bits, which will be described later, are converted into a signal form having a frame structure, and
Error correction encoding processing is performed as necessary. Transmission data is obtained from the framing circuit.

The output signal of the blocking circuit 2 is supplied to a subtraction circuit 5 via a delay circuit 4. The delay circuit 4 delays the output signal of the blocking circuit 2 by the time necessary to detect the maximum value MAX and the minimum value MIN. The minimum value MIN is supplied to the subtraction circuit 5, and data after the minimum value is removed is obtained from the subtraction circuit 5.

The data normalized by this minimum value removal process is supplied to the quantization circuit 7. The dynamic range DR is supplied to the quantization circuit 7, and a 4-bit quantization code DT, for example, is taken out from the quantization circuit 7.

In the quantization circuit 7, the dynamic range DR is (2'-
16) The value of the output data of the subtraction circuit 5 is divided by the quantization step, which is an equal divided value, and the obtained quotient is rounded down to form an integer (!child code DT). The quantization circuit 7 can be realized by a ROM or a division circuit.

The 4-bit data DT from the quantization circuit 7 is supplied to a 2-bit selection circuit 12 that selects the upper 2 bits and a sample delay circuit 17. The upper two bits selected by the two-bit selection circuit 12 are supplied to one input terminal 14 of the switch circuit 13.

The other input terminal 15 of the switch circuit 13 is supplied with a 2-bit output signal from a quantization circuit 21 (described in detail later) that takes visual characteristics into consideration. Input terminal 2 of quantization circuit 21
2 is supplied with the output signal of the blocking circuit 2, the dynamic range DR and the minimum value MfN are supplied from the terminals 23 and 24 to the quantization circuit 21, and the quantization output is generated from the output terminal 25. do. Switch circuit 13
A 2-bit output signal is obtained at the output terminal 16 of.

In this example, the original 8-bit image data is
The data is compressed into 4-bit data, and further compressed into 2-bit data.

The switch circuit 13 is controlled by the output signal of the negative number output circuit 19. - The number output circuit 19 outputs "1" (high level) when two consecutive sample data in the horizontal direction are different and these sample data take a level near the changing point of the 2-bit code. A match signal is generated. In cases other than the above, a signal of "0'" (low level) is generated. - The output signal of the number output circuit 19 is generated for each sample data. - When a failure signal ("1") is generated, the input terminal 15 of the switch circuit 13 is selected, and therefore, the output signal of the quantization circuit 21 that takes visual characteristics into consideration is output. When the coincidence signal is "0", that is, a mismatch signal, the input terminal 14 of the switch circuit 13 is selected, and therefore, the code signal of the upper 2 bits from the 2-bit selection circuit 12 is output.

A sample delay circuit 18 is connected to a sample delay circuit 17 to which the output signal of the ADRC quantization circuit 7 is supplied. Two horizontally consecutive sample data from sample delay circuits 17 and 18 are supplied to a coincidence detection circuit 19. - The number output circuit 19 is supplied with reference data from the ROM 20.

As shown in FIG. 3, from the 4-bit quantization code DT (0000) corresponding to the minimum value MEN of the block to the 4-bit quantization code DT (11) corresponding to the maximum value MAX of the block.
11) is formed by ADRC. The upper two bits (MSB and second MSB) of this 4-bit code are selected by the 2-bit selection circuit 12. Therefore, (00)(01
)(10)(11) 2-bit code signals are formed. In FIG. 3, the 4-bit code marked with a book has values above and below the changing point of the 2-bit code. If the codes of two consecutive samples in the horizontal direction are the codes above and below the change point, deterioration occurs in which a gentle level change turns into a drastic change, as described at the beginning. Therefore, in this case, the output code of the quantization circuit 21 that performs quantization such that the level change is close to that of the original image data is selected.

The ROM 20 generates 8-bit reference data consisting of codes marked with * above and below the change point.

This standard data is as follows.

- In the number output circuit 19, the sample delay circuits 17 and 18
Two samples of data (8 bits) from the above are compared with the six reference data mentioned above. This comparison can be done all at once in parallel, or serially in a time-sharing manner.

b. Quantization circuit that takes into account visual characteristics FIG. 4 shows an example of a quantization circuit 21 that takes into account visual characteristics. The signal is supplied to one input terminal of the selector 30 via the signal. Delay amount D of delay circuit 27
LL corresponds to the time required for the maximum and minimum values to be detected. The delay amount DL2 of the delay circuit 28 corresponds to the horizontal interval between pixels, that is, one sample period.

The output signal of delay circuit 28 is supplied to the other input terminal of selector 30. Output signal of delay circuit 27 and selector 3
The output signal of 0 is supplied to the subtraction circuit 31, and a horizontal difference Δr of the original pixel data (true value) is calculated. When the true value of the pixel of interest is xl and the true value of the pixel one sampling period before is xQ, it is expressed as (Δr-xi-xo).

However, in the case of a pixel in the leftmost column of a block, there is no data for the previous pixel in the block, so the previous block N-
It is necessary to form the difference using the data of the rightmost pixel of 1. The selector 30 selects the data of the rightmost pixel of the previous block from the delay circuit 29 when the leftmost pixel in the block is supplied to the subtraction circuit 31 . Delay circuit 2
The delay tlDL3fi of 9 is set to <(1 block period - 3 sample periods). The selector 30 is controlled by a select signal from a select signal generating circuit 32. The select signal generation circuit 32 is supplied with clock signals (sampling clock and block clock) from the terminal 33, and forms a select signal for controlling the selector 30 as described above.

The true value difference Δr of the image data from the subtraction circuit 31 is supplied to subtraction circuits 34, 35, 36, and 37. The output signals β0, β1, β2, and β3 of these subtraction circuits 34 to 37 are supplied to a minimum value detection circuit 38, and the minimum output signal is detected. The detection signal of the minimum value detection circuit 38 is supplied to the code selection circuit 39, and a 2-bit code signal is generated from the code selection circuit 39. The code signal is taken out to the output terminal 25. The code selection circuit 39 selects one of the 2-bit code signals (00) (01) (10) (11) corresponding to the decoding representative levels 10111, I2, and I3, respectively.

The selection operation of the code selection circuit 39 is as follows.

If β0 is the minimum, (00) is selected.

If β1 is the minimum, (01) is selected.

If β2 is the minimum, (10) is selected.

If β3 is the minimum, (11) is selected.

For subtraction circuits 34 to 37, subtraction circuits 40.41.4
Signals Δ0, Δ1, Δ2, Δ3 are supplied from 2.43. These signals Δ0 to Δ3 are the differences between the decoding level (XO) of the pixel before the pixel of interest and the four representative decoding levels, and are predicted changes. The subtraction circuits 34 to 37 and the minimum value detection circuit 38 detect the one closest to the horizontal difference Δr of the true value of the image data among the signals Δ0 to Δ3. That is, the code signal corresponding to the decoding representative level that causes the change closest to the horizontal signal change of the original image signal is selected for the pixel of interest.

The subtraction circuits 40 to 43 include local decoders 48 and 49.
．． The decoding representative level formed by 50.51 (MIN+1
0)(MIN+11)(MIN+12)(MIN+I3
) is supplied. In order to generate these decoding representative levels, the dynamic range DR and the minimum value MIN are supplied to the local decoders 48 to 51, and the terminal 4 is supplied with the dynamic range DR and the minimum value MIN.
4.45.46.47 to 2-bit code signal (00
)(01)(10)(11) are supplied, respectively.

The local decoders 48 to 51 and 52 are each configured with a ROM to which the dynamic range DR and the quantization code DT are supplied as addresses, and a minimum value MIN is added to the data read from the ROM.

The decoding level XO of the pixel before the pixel of interest is formed by the local decoder 52, delay circuits 53, 54, 55, and selector 55. The local decoder 52 is supplied with a code signal from the code selection circuit 39, and the local decoder 52 generates a decoding level of the pixel of interest. This decoded level is supplied to one input terminal of the selector 55 via a delay circuit 53 having a delay amount DL2 of one sampling period. The output signal of the delay circuit 53 is supplied to the other input terminal of the selector 55 via the delay circuit 54 having a delay amount DL3 of (1 block period - 3 sampling periods). The selector 55 is controlled by a select signal from the select signal generation circuit 32, similar to the selector 30 described above.

The delay circuits 53 and 54 and the selector 55 generate the decoding level xO of the pixel XO before the pixel of interest x1 in the same manner as the delay circuits 28 and 29 and the selector 30 described above. This decoding level is supplied to subtraction circuits 40-43. Therefore, the signals ΔO to Δ3 generated from the subtraction circuits 40 to 43, respectively.
is the predicted difference between the four decoded representative levels and the previous pixel's decoded level XO, as follows:

Δ0= (IO+MIN)-XO Δ1= (11+MIN) -XO Δ2- (12+MIN) -XO Δ3- (130 MIN) -XO In the subtraction circuits 34 to 37, β0-Δr-ΔO β1-Δr-Δl β2-Δr An output signal of −Δ2 β3−Δr−Δ3 is formed. By the minimum value detection circuit 38,
Since the smallest one among βO to β3 is detected, the code selection circuit 39 selects the 2-bit code whose predicted change amount is closest to the true change amount Δr.

It is also effective to perform quantization so as to faithfully represent not only the horizontal difference in the above embodiment but also two-dimensional changes in the horizontal direction, vertical direction, diagonal direction, and the like. For example, as shown in FIG. 5, the level of the pixel of interest is X, and the levels of surrounding pixels above, to the left, and diagonally above are a, respectively.
b, c, the amount of change Δr and the predicted amount of change Δi (i-0, 1, 2, 3) in the true value of the pixel of interest are calculated as the difference between the average level of the surrounding pixels and the level of the pixel of interest. It will be done. That is, Δr - (3x-a-b-c) Δt - (31i -A-B-C) (However, A, B, and C are decoded quantization codes obtained for a, b, and c, respectively. level.

The quantization code signal that yields Δl closest to Δr is selected.

In order to obtain the amount of change in space, the predicted value obtained by spatial prediction may be used instead of the average value. In other words, if the predicted value is X', then x''mb+%(c -a) Δr-x-x -x-b-'74 (c-a) Δim
I t-B-% (C-A) Then, similarly to the above, the quantization code signal that yields Δl closest to Δr is selected.

C0 Modification In the above-described embodiment, the quantization circuit 21 performs quantization so that the spatial changes become close to the original image data after decoding. However, two temporally consecutive sample data at spatially corresponding positions, for example two sample data with a time difference of one frame, are quantized so that the change between them is close to the change in the true value. A quantization circuit may also be used.

Further, the present invention can be applied to quantization circuits for other high efficiency codes such as variable length ADRC and three-dimensional block ADRC.

[Effects of the Invention] According to the present invention, it is possible to prevent visually noticeable deterioration in which a gentle change in original image data turns into a violent change after decoding. Moreover, in this invention, quantization is performed to preserve changes in the original image signal only when there is a risk of the above-mentioned degradation occurring; in other cases, quantization is performed based on level, so that the original image signal and level quantization with good correspondence can be performed.

[Brief explanation of the drawing]

FIG. 1 is an overall block diagram of one embodiment of the present invention, and FIG.
The figure is a route map of an example of a block, Figure 3 is a route map used to explain the process of compressing a 4-bit code to 2 bits, Figure 4 is a block diagram of an example of a quantization circuit that takes visual characteristics into consideration, FIG. 5 is a route map used to explain another example of spatial change detection, and FIGS. 6 and 7 are route maps used to explain conventional quantization. 19F-number construction circuit, 31; subtraction circuit that generates true value variation, 388 minimum value detection circuit, 39: code selection circuit, 40.41.42.438 subtraction circuit that generates predicted variation, 48. 49.50.51,52: Local decoder. Agent Patent Attorney Masato Sugiura Explanation of main symbols in the drawing 1: Digital video signal input terminal, 7: ADRC quantization circuit, 12: Circuit for selecting upper 2 bits, 13: Switch circuit, 16: Output Terminal, −Real! Sail example Figure 1 Figure 1 + Nootsu Figure 3

Claims (1)

[Claims] Digital data quantized with n bits is converted into m bits (
In a quantization circuit for image data that is compressed into data (n>m), the codes of the upper m bits of the above n-bit digital data are arranged between two spatially adjacent sample data or spatially correspond to each other and are temporally continuous. A detection circuit detects that there is a difference between the two sample data and that both of these two sample data take a level near the changing point of the m-bit code, the true value of the pixel of interest, and the true value of the pixel of interest and the surroundings of the pixel of interest. or between the true value of the pixel of interest and the true value of a reference pixel that temporally precedes and spatially corresponds to the pixel of interest; Detect the amount of temporal change in the decoded value of the quantization code, and determine the decoded representative value for which the amount of spatial or temporal change in the reference pixel with respect to the decoded value of the quantization code is closest to the detected amount of spatial or temporal change. seek,
a quantization circuit configured to select a quantization code corresponding to the decoded representative value; and m-bit digital data formed from the n-bit digital data and the output of the quantization circuit to the detection circuit. 1. A quantization circuit for image data, comprising a selection circuit that selectively outputs an output signal according to an output signal.