JPH01227590A - System for compressing and expanding digital image data - Google Patents

Abstract

PURPOSE:To easily obtain the reproducing image of a satisfactory picture quality by using a rectangular window function as a two-dimensional window function for the signal processing of digital data at the time of executing data compression and expansion. CONSTITUTION:As the window function, a function, in which a weighting to the signal of an overlapping part between respective block images in a part corresponding to the overlapping area part of respective blocks to be adjoining is made into 1 by generalizing the signal processing both at the time of compressing the data and the time of expanding data, is used. To the digital data to which the compression of the data is executed by a Fourier transform, the window function is applied. Further, also to the digital data to which the expansion of the data is executed by the inversion of the time of compressing the data, the window function is applied, and the digital data corresponding to the part of overlapping area to be specified beforehand between respective adjoining blocks are added. Thus, the continuity of a signal waveform between respective adjoining block images can be made satisfactory, and the reproducing image with the satisfactory picture quality can be easily obtained.

Description

[Detailed description of the invention] (Industrial application field) The present invention relates to the recording and reproduction of image information and the transmission of image information.

This invention relates to a compression/expansion method for digital image data that can be applied to transmission, reception, etc.

(Prior art) When digitizing an image signal, linear quantization (equal quantization) is used to equally divide the signal level for each sample value and replace each range number with one representative value. ), and in order to make the difference between the representative point and the original value indiscernible, 6 bits (3? tones) to 8 pits (256 tones) are generally required for natural images. Therefore, if we try to record the signal that has been digitized by uniform quantization of the image signal as described above, it is necessary to handle a large amount of information as described above for each sample value. be done.

Therefore, even though the signal is encoded with a smaller amount of information, the parts where the signal changes little are sensitive to changes, and the parts where the signal changes rapidly are sensitive to changes, even if there is some error, it cannot be detected. The amount of information per sample was reduced by taking advantage of the difficult nature of human vision and hearing, or by using the correlation on the time axis in the information signals being recorded. After recording, transmitting, and transmitting digital data whose data amount has been compressed by applying various high-efficiency encoding methods, and after playing and receiving digital data whose data amount has been compressed as described above, , it has been conventional practice to decompress data.

FIG. 10 shows an example configuration of a conventional digital image data compression/expansion system configured to compress digital image data and transmit/receive it, and to output the received digital image data as an analog image signal after data expansion. In this block diagram, the analog image signal to be transmitted and received (the same applies to the image signal to be recorded and reproduced) is converted by an analog-to-digital converter (ADC) 1. After being converted into a digital signal, the image data for each frame period is transferred to the memory 1 indicated by reference numeral 3 in the drawing and the memory 1 shown in the drawing. The data are sequentially and alternately written and stored in the memory 2 indicated by reference numeral 4.

When one of the memories 1 and 2 is in the write mode, the other is in the read mode. When digital data is being supplied to one memory that is in the write mode via the changeover switch 2, the digital data read from the other memory that is in the read mode is supplied to the changeover switch 5. It is designed to be supplied to subsequent circuits via.

The mode of operation as described above by the changeover switches 2 and 5 in FIG. 10 and the memories 1 and 2 indicated by the drawing reference numeral 3.4 will be explained later by the drawing reference numeral 3. The same applies to the operation mode of the memory 1゜2 and the changeover switch 2.5 indicated by 4, and the memory 3゜memory indicated by the drawing reference numeral 12.13 in FIGS. 1 and 10. 4 and the changeover switch 11.14 as well as the memory 5. The same applies to the operation mode of the memory 6 and the changeover switches 16 and 19.

In the conventional digital image data compression/expansion method shown in FIG. 10, the digital data output from the memory 1.2 indicated by reference numerals 3 and 4 in FIG. The data is supplied to a converter 6 for data compression.

Compression of digital data in the orthogonal transformer 6 is performed, for example, by dividing the image data of one frame period into a plurality of blocks (for example, 16 pixels x 16 pixels/block) and converting the image data of each block into orthogonal data. Transformation (Fourier transform, Hadamard transform, cosine transform) is performed, and the two-dimensional frequency component is compressed by requantization or truncation in the vector quantizer 7, and then supplied to the modulator 8. Ru.

Then, the digital data compressed as described above is sent to the modulator 8, where it is modulated by a predetermined modulation method and then recorded or transmitted.

The digital data compressed as described above is demodulated by a demodulator 9 and then sent to an inverse vector quantizer 10.
Inverse vector quantization is performed at .

Digital data subjected to inverse vector quantization processing in the inverse vector quantizer 10 is transferred to a changeover switch 11 and a memory 3 indicated by a reference numeral 12 in the drawing, and a memory 4 and a changeover switch 14 indicated by a reference numeral 13 in the drawing. The digital data is sequentially and alternately written into the two memories in the circuit arrangement and sequentially and alternately read out, and the read digital data is supplied to the inverse orthogonal transformer 15.

The inverse orthogonal transformer 15 performs an inverse orthogonal transform process on the digital data supplied thereto to expand the data,
Data is sequentially and alternately written into two memories in a circuit arrangement consisting of a changeover switch 16 and a memory 5 indicated by a reference numeral 17 in the drawing, and a memory 6 and a changeover switch 19 indicated by a reference numeral 18 in the drawing. The read digital data is read out to the digital-to-analog converter 20.
is supplied to The digital-to-analog converter 20 performs digital-to-analog conversion on the digital data supplied thereto, and outputs an image signal in the form of an analog signal.

Figure 11 is a timing chart of the recording (or transmission system), and Figure 12 is a timing chart of the reproduction (or reception) system, where WT represents writing and RD represents reading each time. .

In the conventional digital image data compression/expansion method shown in FIG. 10, the digital data is subjected to orthogonal transformation and inverse orthogonal transformation as signal processing for data compression and data expansion. Image reproducibility in such cases deteriorates due to the presence of aliasing distortion, quantization errors, conversion calculation errors, etc. However, slight aliasing errors are tolerated, and quantization errors are within a range that is not visually noticeable. By suppressing this (eliminating waveform discontinuity), approximately good image reproduction can be achieved.

(Problems to be Solved by the Invention) By the way, in the conventional digital image data compression/expansion method described above, orthogonal transformation is applied to digital data as a signal processing means for compressing/expanding digital data. Inverse orthogonal transformation refers to converting one image (one screen) targeted for data compression as shown in Figure 13.
Divided by Nv in the vertical direction and Nh in the horizontal direction (NvXNh)
High-efficiency encoding and decoding were performed by applying this method to each block image in the block images. FIG. 14 shows a specific example of one of the (NvXNh) block images shown in FIG. 13. Therefore, one block image shown in FIG. This is an example of a case where it is configured.

FIG. 15 shows an example in which orthogonal transformation (for example, Fourier transformation) is performed on the image data of one block out of a plurality of blocks of image data of one frame period. When a Fourier transform is performed on a certain block image, there is a tendency for the low-frequency components to be large and the high-frequency components to be small. The sizes of the circles in FIG. 15 indicate the levels of the frequency components. Note that when the image is a color image, it goes without saying that each frequency component also exists for each primary color.

Now, when digitizing an image signal, in order to make the difference between the representative point and the original value indiscernible by using linear quantization (uniform quantization), it is generally necessary to use 6-bit ( As mentioned above, it is said that 8 bits (25,656 gradations) are required from 32 gradations), but conventionally, data compression when the gradation of the original pixel was 8 bits was difficult, for example. Fine quantization (e.g., 8 bits) is applied to the high-frequency components, and coarse quantization (e.g., 4 bits) is applied to the high-frequency components. ingredients and
This is done by, for example, cutting off signal components between frequency values fv2 to fv1.

However, when data compression/expansion as described above is performed, requantization is performed after orthogonal transformation.

or data is truncated, and signal processing is performed for each block image, which is equivalent to using a rectangular window function as a two-dimensional window function for signal processing of digital data. , errors occur due to aliased frequency components. For the reasons described above, a problem arises in that waveform discontinuities occur at the boundaries of signals corresponding to each block image obtained by inverse transformation during reproduction, resulting in deterioration of image quality.

The influence of the rectangular window function mentioned above and problems such as discontinuity of reproduced waveforms are explained as follows with reference to waveform diagrams and other figures. (Explain using signals as an example).

FIG. 16 is a diagram showing that a one-dimensional continuous signal is divided into block signals at fixed time lengths on the time axis by a rectangular window, and FIG. 17 is similar to the above-mentioned FIG. This shows a signal waveform that is treated mathematically when the one-dimensional signal waveform shown is made into a block signal every fixed time length by a rectangular window.

As is well known, the frequency spectrum of a continuous sine wave signal as shown in FIG. 16 is as shown in FIG. 18 because it has only a single frequency component. The spectrum of the signal shown in FIG. 17, which has been made into a discontinuous wave by the above-mentioned method, is broadened as shown in FIG. 19.

In addition, Fig. 20 shows that when the signal targeted for sampling has only frequency components that are less than half of the sampling frequency fs, aliasing distortion does not occur, and the frequency components that are less than half of the sampling frequency fs are passed through at low frequencies. It is shown that by extracting with a filter, a reproduced signal in an undistorted state can be obtained, and Fig. 21 shows that the signal targeted for sampling also includes frequency components that are more than half the sampling frequency fs. In this case, aliasing distortion occurs and a distortion-free reproduced signal cannot be obtained.

Due to the above-mentioned distortion, requantization after orthogonal transformation during data compression, and truncation of some components, waveform discontinuity occurs between adjacent blocks during playback, and aliasing distortion occurs in the playback image. causes moiré. FIG. 22 shows an example of the original signal waveform before data compression, and FIG. 23 shows an example of the signal waveform after data compression and expansion of the signal shown in FIG. al, a2. The part of the signal indicated by a3 and the 23rd
a 1 j , a 2 j , a 3 in the diagram
The signal portion indicated by j is the corresponding signal portion.

The discontinuity of the waveform causes a mosaic pattern on the reproduced image in which the brightness and color differ slightly for each block image.

(Means for Solving the Problems) The present invention provides a predetermined overlapping area between adjacent block images, and data compression is performed on the digital data of the block images by Fourier transformation. As a window function that gives weight to the previous digital data and the digital data after data decompression is performed by inverse transformation to that during 2-data compression, the The weighting of the signal of the overlapped part between each block image is calculated by integrating the signal processing during both data compression and data expansion.
The above window function was applied to the digital data that should be compressed by Fourier transform using what should be done, and the data was decompressed by the inverse transformation to that used when compressing the data. A digital device comprising means for applying the window function to digital data as well, and means for adding digital data corresponding to a predetermined overlapping area between adjacent block images. Provides a compression/expansion method for image data.

(Example) Hereinafter, specific contents of the digital image data compression/expansion method of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 shows an example of a compression/expansion method for digital image data according to the present invention, which is configured to compress digital image data and transmit/receive it, and to expand the received digital image data and then output it as an analog image signal. 1 is a block diagram showing an embodiment, and in this FIG. C) After being converted into a digital signal by 1, the image data for each frame period is indicated by reference numeral 3 in the drawing by a changeover switch 2 which performs a switching operation using one frame period in the image signal □ as a switching cycle. The information is sequentially and alternately written and stored in the memory 1 shown in FIG.

When one of the memories 1 and 2 is in the write mode, the other is in the read mode. When digital data is being supplied to one memory that is in the write mode via the changeover switch 2, the digital data read from the other memory that is in the read mode and is
The data is supplied to subsequent circuits via a changeover switch 5.

The mode of operation as described above by the changeover switches 2, 5 and the memories 1, 2 indicated by the drawing symbols 3 and 4 in FIG. 1 is indicated by the drawing symbols 12 and 13 in FIG. Memory 3. The operation of the memory 4 and the changeover switch 11.14 as well as the memory 5.1, which is designated by the drawing reference numeral 17.18. The same applies to the operation mode of the memory 6 and the changeover switches 16 and 19.

In the compression/expansion method for digital image data shown in the first diagram, memories 1 and 2 are designated by reference numerals 3 and 4 in the diagram.
The digital data outputted from the switch 5 is supplied to a two-dimensional window function multiplier 21, and the two-dimensional window function multiplier 21 uses a window function as shown in FIG. Weighting is applied to the signal shown as the original waveform in FIG. 25, and a signal having the waveform shown in FIG. 26 is output.

FIG. 24 shows sequential blocks bl, b2 . b
3 schematically shows the overlapping areas of the blocks bl, b2, . 25 is a diagram schematically showing the overlapping region of b3, and as described above, the signal shown as the original waveform in FIG. 25 indicates a signal portion corresponding to one block image portion in FIG. 24. ing.

A predetermined overlapping area is provided between each adjacent block image.The digital data before data compression is performed by Fourier transform on the digital data of each block image described above, and the digital data at the time of data compression. The aforementioned two-dimensional window function multiplier 2 is used to weight the digital data after data expansion through inverse transformation.
As a window function the digital data should be multiplied by 1.

The weighting of the signal of the overlapped portion between each block image in the overlapped region portion of each adjacent block and the corresponding portion is made 1 by integrating the signal processing during both data compression and data expansion. is used.

The digital data weighted by the above-mentioned two-dimensional window function multiplication @21 and the window function is supplied to the orthogonal transformer 6. The orthogonal transformer 6 divides the image data of one frame period into a plurality of blocks, performs Fourier transform on the image data of each block, and then converts the digital data compressed by the orthogonal transform by the orthogonal transformer 6. The two-dimensional frequency component is further compressed by requantization or truncation in the vector quantizer 7, and then supplied to the modulator 8.

The digital data compressed as described above is modulated by a predetermined modulation method in the modulator 8 and then recorded or transmitted.

The digital data compressed as described above is demodulated by a demodulator 9 and then sent to an inverse vector quantizer 10.
Inverse vector quantization is performed at .

Digital data subjected to inverse vector quantization processing in the inverse vector quantizer 10 is transferred to a changeover switch 11 and a memory 3 indicated by a reference numeral 12 in the drawing, and a memory 4 and a changeover switch 14 indicated by a reference numeral 13 in the drawing. The digital data is sequentially and alternately written into the two memories in the circuit arrangement and sequentially and alternately read out, and the read digital data is supplied to the inverse orthogonal transformer 15.

The inverse orthogonal transformer 15 performs an inverse orthogonal transform process on the digital data supplied thereto to expand the data. FIG. 28 schematically shows the output signal from the inverse orthogonal transformer 15, and the output signal from the inverse orthogonal transformer 15 is supplied to the two-dimensional window function multiplier 22.
After being weighted by the window function in the two-dimensional window function multiplication @22, the signal is supplied to the overlap section adder 23.

The overlap section addition section 23 performs an operation of adding a predetermined overlap region set between adjacent block images and a corresponding signal portion. The signals in the overlapping area are added by reading out the signals in a predetermined overlapping area in each adjacent block stored in the buffer memory, and then adding the signals in the overlapping area using a digital signal that can generate a signal as shown in FIG. Data is output.

The output signal from the overlap section adder 23 is transferred to a selector switch 16, a memory 5 indicated by reference numeral 17, a memory 6 indicated by reference numeral 18, and a selector switch 19.
The digital data is sequentially and alternately written into the two memories in the circuit arrangement and read out sequentially and alternately, and the read digital data is supplied to the digital-to-analog converter 20.

The digital-to-analog converter 20 performs digital-to-analog conversion on the digital data supplied thereto, and outputs an image signal in the form of an analog signal.

FIG. 2 is a timing chart of the recording (or transmission system), and FIG. 3 is a timing chart of the reproduction (or reception) system. At each time, WT writes, R
D represents reading.

In the digital image data compression/expansion method of the present invention shown in the first diagram, Fourier transform is applied to the digital data of each block image described above in which a predetermined overlapping area is provided between adjacent block images. The above two-dimensional window function multipliers 21 and 22 are used to weight the digital data before data compression is performed by the above-described method and the digital data after data expansion is performed by inverse transformation to that during data compression. As a window function to be multiplied by the digital data by 27. Since the signal processing that can be integrated into 1 is used, the spread of the spectrum is improved compared to that shown in FIG. 19 in the conventional example described above. The above window function is applied to digital data that should have little aliasing distortion, and should be compressed by Fourier transform, as shown in the figure. also applies the window function described above to the digital data that has been expanded by inverse transformation, and then adds the digital data corresponding to the predetermined overlapping area between each adjacent block image. Therefore, the continuity of the signal waveform between adjacent block images is good, and a reproduced image with good image quality can be easily obtained.

Next, an example of a two-dimensional window function will be explained with reference to FIG. 4. In FIG. 4, which shows a portion of a plurality of adjacently arranged block images, it is shown that a predetermined overlapping area is provided between each adjacent block image.

Block image B(i+1) now shown in FIG.
, j, areas (A), (C), CG), and (I) in block image B(i+1)*j are four block images adjacent to each of the four sides of block image B(i+ILj). This overlap area is weighted by a window function in both the X and Y directions.

Also, block image B(i+1), area (B) in j
.

CD), (F), and (H) are block images B(i+1)
, and two of the four block images adjacent to each of the four sides of j, and this overlapping area only exists in either the X direction or the Y direction. are weighted using a window function.

Furthermore, the area (
E) is an area where pixels do not overlap with any of the four block images adjacent to the four sides of block images B(i+1) and j, and the weighting by the window function is 1.
This is an area that is considered to be

FIG. 4 shows the digital image data of each block image described above in which a predetermined overlapping area is provided between adjacent block images in the digital image data compression and expansion method of the present invention shown in FIG. The two-dimensional window function multiplier described above is used to weight the digital data before data compression is performed by Fourier transform and the digital data after data expansion is performed by a transform opposite to that at the time of data compression. 21. As the window function to be multiplied by the digital data in 22, the weighting of the signal of the overlapped part between each block image in the part corresponding to the overlapped area part of each adjacent block is determined at the time of data compression and the time of data expansion. This is an example of a case where weighting of an overlapping area with an adjacent block is performed using a sine function so that the signal processing by both blocks is combined to be 1.

Therefore, the block image B(i+1) is obtained through the entire digital image data compression/expansion system shown in FIG.

The overall weighting performed for area (A) in j is: B(i+1), component of j = (sin(πX/2・Q2) sin(πy/2・m2
))” =41) Bi, component of j == [5in(x (Q 2-X)/2・It 2)
sin(x y/2・w+2)]”=(cos(πX/
2・n 2) sin(πy/2・sao2))2...
(2) Components of B (i+1), (j +1) - [5in(πX/:”Q 2)sin(π(++
+2-y)/2・m2))ko2=(sin(cX/2・
Q2) cos(πy/2・m2))” −(3) Bi
, component of (j +1) = [5in(tc (n 2-X)/2・Q 2)s
in(π(Q 2-y)/2・m2)]”=(cos(
πX/2・Q2)cos(πy/:'+a2))"
-(4) Kokote, (πX/2・Ω2) =X, hcy
/2・m2)=Y, and when calculating the sum Ta of the four components shown in equations (1) to (4) above, Ta = sin"X-5in"Y +cos"X- 5
in”Y +sin”X・cos' Y + cos”
X @cos” Y=(sun”Xll5in”X)
(sin"Y + cos"Y) = 1-(5) The above (
As shown in equation 5), block images B(i+1),
The overall weighting performed for region (A) in j is 1.

The area (
A) and areas (C), (G), (I), etc. are under the same conditions, so throughout the digital image data compression/expansion system shown in Figure 1, the areas in block images B(i+1), j It goes without saying that the overall weighting performed on (C), CG), and (I) is also 1.

Next, the overall weighting performed on the area (B) in the block image B(i+1),j throughout the digital image data compression/expansion system shown in FIG. 1 is as follows: Component of B(i+1),j = ( sin(zy/2・l12))' =-(6)B
Components of (i+1), (j +1) = [5in(7C(m2-y)/2・m2))]”=
(cos(πy/2・m2))” −(7) Here,
(6) above, assuming (πy/2·m2)=Y.

When calculating the sum Tb of the two components shown in equation (7), T b =-sin"Y +cos"Y = 1-
(8) As shown in equation (8) above, the overall weighting performed on the area (B) in block image B (i+ 1), j throughout the digital image data compression/expansion system shown in Figure 1 is It becomes 1.

Furthermore, block images B(i+1), j
The overall weighting performed on the region (D) at
B (1+ 1) component of v j = [5in(tc (112-y)/2・Q 2))
]”=(cos(πy/2・m2))” −(10)
Here, (zx/2・Q 2)=X, (zy/2
・M2)=Y, and when calculating the sum Tb of the two components shown in equations (9) and (10) above, Td=sin"X +cos"Y=1-(11
) As shown in equation (11) above, block image B is
(i+1) The overall weighting performed on region (D) at tj is 1.

In this way, when the weighting of the signal of the overlapped portion between each block image in the portion corresponding to the overlapped region portion of each adjacent block image is performed using a sine function,
The signal processing performed during both data compression and data expansion is combined into one, and the original image is correctly reproduced.

Note that although the overall weighting for the edges of the image is not 1, the problem of reproducibility can be solved by applying means such as adding black level dummy pixels to the edges, for example.

The weighting of the signal of the overlapped portion between each block image in the overlapped region portion of each adjacent block and the corresponding portion is made 1 by integrating the signal processing during both data compression and data expansion. In addition to the above-mentioned sine function, the weighting mode to be applied to the overlapping area with adjacent blocks may be a function as shown below, for example.

That is, the two-dimensional window function f (x*y) e g h,
y) are the same, O≦X≦Q1+Q2, 0≦y≦m
At l+m2, (f77""FT) (fyl;T) :X≦Q 2
．． 76m2 (f7711): x≦Q2
．． m2≦y≦m1 (fv71d): m
2≦y≦at, 76m2゜1: Q2
≦X≦Ill, m2≦y≦m1 (ml+m2−y/m2
): f12≦X≦Q 1.72m1 (QL+Q2-X)
/E12): x≧Ql, m2≦y≦ml.

(Q1+ff12-x)/Q2) (ml+m2-y)/
m2): X≦11t, 75m1 (f771 completed) (ml+m2-y)/m2):X≦Ω2
,75m1 (121+12-:c)/j! 2) (v”;)'m2)
, :X≧ax, 76m2 Function f (x+y)t g (
x, y).

FIG. 5 shows one block image, and each pixel (Pij) in the case of a color image consists of a plurality of elements, such as a luminance component and two color difference components, or three primary color components.

When a color image is represented by a luminance component and two color difference components, the color difference component may have a smaller amount of information than the luminance component, as is well known.

For example, only information such as the odd numbered item in the first row, the even numbered item in the second row, the odd numbered item in the third row, etc. is considered to be valid information, and the amount of data is halved compared to the amount of luminance information. You can. In this case, it is reasonable to double the block size of the color difference signal both horizontally and vertically because the amount of information becomes the same as that of the luminance signal and signal processing can be performed in the same way.

Luminance elements: Pll, Pl2. ...P21. P2
2... Color difference element-1: Pll, Pl3. ...P2
2. P24. ...P31. P33 color difference element-2: P
l2. Pl4. ...P21. P23. ...P32.
P34 By doing this, two frames of luminance memory can constitute one frame of color image memory.

Figure 6 is a diagram in which weighting by a two-dimensional window function is used to explain addition, and in this example, both images of one block are 1
It is assumed that the image consists of 6×16 pixels, and the overlapping area with adjacent block images is 4 bits in each direction.

It is assumed that the window function applied to the overlapping area of each block image is a sine function, the lower left pixel in FIG. 6 is set to Pl,1, and the upper right pixel in FIG. 6 is set to Pl6. lf
As i, the weighting of each pixel in the one block image shown in FIG. 6 is shown as follows: P11' to P16,16', respectively.

Pi, 1'=O,...P 1,16'=
OF2,1'=O,P2,2'=sin(π/8)s
in(π/8)P2, 2P 2.3' = 5in(π
/8) sin(2tc /8) P 2,3P 2.4
' = 5in (π/8) sin (3n /8) P
2e4P2,5'=sin(s/8)P2,5-...-
... FIG. 7 shows a state in which pixel data of one block image shown on the left side is orthogonally transformed into two-dimensional Fourier transformed data shown on the right side of FIG. In reality, high-speed calculations are possible using fast Fourier transform.

The diagram shown on the left side of FIG. 8 is a diagram of the two-dimensional Fourier transform data shown on the right side of FIG. 7, and the diagram shown on the right side of FIG. The two-dimensional Fourier transform data with sufficient accuracy shown on the right side is
This shows the state of vector quantization for data compression.

In general, low frequency components (lower left Fo, close to Fo) are finely quantized (quantized with a large number of bits), and high frequency components are coarsely quantized (quantized with a small number of bits). , and even truncates the upper right frequency component.

FIG. 9 is an explanatory diagram of the case where reproduced pixel data is obtained by inverse Fourier transform from data compressed as described above. is also possible.

(Effects of the Invention) As is clear from the above detailed explanation, the compression/expansion method for digital image data of the present invention provides a predetermined overlapping area between each adjacent block image. A window function that weights the digital data of a block image before data compression is performed by Fourier transform, and the digital data after data expansion is performed by inverse transformation to that during data compression. As such, the weighting of the signal of the overlapped part between each block image in the part corresponding to the overlapped area of each adjacent block should be made 1 by integrating the signal processing during both data compression and data expansion. The window function is applied to the digital data to be compressed by Fourier transform using the same method.

Means for applying the window function to digital data which has been expanded by inverse transformation compared to data compression, and a predetermined overlapping area between adjacent block images. A compression/expansion method for digital image data, comprising means for adding corresponding digital data, and a means for adding corresponding digital data. Since the signal processing of digital data was equivalent to using a rectangular window function as a two-dimensional window function, errors due to folded frequency components occurred, and each block image obtained by inverse transformation during playback According to the present invention, a reproduced image of good image quality can be easily obtained without causing the problem of deterioration of image quality due to waveform discontinuity at the boundary of the corresponding signal.

[Brief explanation of the drawing]

FIG. 1 shows an example of a compression/expansion method for digital image data according to the present invention, which is configured to compress digital image data and transmit/receive it, and to expand the received digital image data and then output it as an analog image signal. A block diagram showing the embodiment, FIGS. 2 and 11 are timing charts of the recording (or transmission system), FIGS. 3 and 12 are timing charts of the reproduction (or reception) system, and FIG.
Figures 5 to 5 are diagrams showing some of a plurality of block images arranged adjacent to each other, and also showing that a predetermined overlapping area is provided between each adjacent block image. Figures 9 and 13 to 29 are diagrams for explaining the operation, and Figure 10 shows how digital image data is compressed and transmitted/received, and the received digital image data is expanded and output as an analog image signal. FIG. 2 is a block diagram of a conventional digital image data compression/expansion method configured in a continuous manner. 1...Analog-to-digital converter (ADC), 2゜5
．． 11, 14, 16. 19... Selector switch, 3゜4
．． 12, 13, 17, 18... memory, 6... orthogonal transformer, 7... vector quantizer-8... modulator,
9--Demodulator, 10--Inverse vector quantizer, 15
... Inverse orthogonal transformer, 21.22... Two-dimensional window function multiplier, 23... Overlap section addition section, 20...
Digital to analog converter, -QQ- Search]60

Claims (1)

[Claims] A predetermined overlapping area is provided between each adjacent block image, and the digital data of the block image is divided into two types: As a window function that weights the digital data after data has been expanded by conversion, the weighting of the signal of the overlapped area between each block image in the area corresponding to the overlapped area of each adjacent block is used for data compression. Applying the window function to the digital data to be compressed by Fourier transform, using the signal processing that should be made into 1 by combining the signal processing during both time and data expansion, and Means for applying the window function to digital data which has been expanded by inverse transformation compared to data compression, and a predetermined overlapping area between adjacent block images. and means for adding corresponding digital data.