JP2916719B2 - Encoding method, encoding device, and encoding / decoding device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding method, an encoding device, and an encoding / decoding device for compressing and encoding a digital video signal.

[0002]

2. Description of the Related Art Conventionally, as a method for encoding a digital video signal, Japanese Patent Application Laid-Open No. 1-253382 and US Pat.
There was one shown in 394,774. Hereinafter, these publications will be described as examples.

The transmission or recording of a video signal in digital form has the potential to significantly reduce the effect of channel noise or read noise on the quality of the displayed image.
It also offers the possibility of easily transmitting these digital signals over telephone-type digital networks. Nevertheless, the digitization of the sequence of television pictures takes place at a very high rate, so that digitized color television signals cannot generally be transmitted or recorded directly on existing carriers. According to CCIR Notice 601, the digitization rate of color television signals is 216 Mbit / s. Therefore, to adapt the digitized color television signal to the actual transmission and recording speed, it is important to reduce this speed.

US Pat. No. 4,394,774 describes a method by which this speed can be reduced by a factor of 10 to 20, ie 1/10 to 1/20. This method, which is based on the use of an orthogonal transform, offers the potential of using the redundancy between neighboring pixels in the video to gain the benefit of compressing that redundancy. This method is characterized in that an image is divided into blocks having the same size, and an orthogonal transform having a property of de-correlating the pixels of the block by concentrating energy on a small number of pixels is performed on each block. I have.

[0005] This method is often combined with inter-frame prediction techniques to obtain the same information compression benefit from the video-to-video redundancy that is present in the still parts of the video. According to this technique, the block itself is transmitted (intra-frame mode) or the difference between this block and, after encoding and decoding, the block having the same spatial position as the preceding video is transmitted ( Either the block with the lowest energy is transmitted.

This video-to-video prediction operation thus introduces the time-recursion of the encoding operation, in other words, if the decoded preceding video is available, the prediction operation states that It is only possible to decode the video signal in normal playback mode. This feature means that errors appearing in the reception or during the reading of the bands are present in the various pictures, and in fact, as long as the block in which this error appears is encoded by the interframe mode, there is a risk that the errors appear in the pictures. The result is that it exists.

In addition, this recursiveness has been observed in consumer video recordings,
That is, it is incompatible with recording by a home video recorder.
This is because random access to video is excluded, and this random access is necessary for realizing the “quick search mode”. In some cases, a remedy for this drawback is to encode one video out of N videos in intra-frame mode, but since this degrades the quality of the displayed video, N limits this degradation. Must be so large that this limits the scope of improvement.

Japanese Patent Laid-Open Publication No. 1-253382 provides a video signal encoding method that can benefit from video-to-video correlation without introducing the video-to-video recursion. That is, this publication provides a method that is compatible with consumer video recording and is not sensitive to channel errors.

The following preliminary steps for realizing the encoding method for this purpose are as follows: (a) Evaluation of the main motion from video to video to associate the displacement vector with each video with respect to the preceding video. And (b) dividing the sequence of video signals corresponding to the images into groups each corresponding to N consecutive images, and Within each group, define M lines of N successive planes corresponding to the image planes of the group on the one hand and N images of the group on the other hand and a three-dimensional block containing P pixels per line. Is a scan conversion for defining a format of a three-dimensional block by performing M-line and P-pixels constituting each three-dimensional block of the same group. N number of two-dimensional blocks by is characterized in that it comprises a preliminary step, which is spatially shifted over the next image from one image by the displacement vectors were evaluated for each video.

The method provided offers the possibility to use the temporal redundancy of the signal, thanks to the inverse correlation realized by the orthogonal transformation of the stationary part of the image without substantial displacement, which Or even in the case of general movement of the camera, and even when the movement affects a large part of the scene. In the last two cases, the provided method is superior to the method using the inter-frame mode and the intra-frame mode. This is because it uses interframe correlation and at the same time the interframe / intraframe process does not consider these displacements. In addition, since the effect is limited to N videos, this process does not introduce any video-to-video recursion during encoding, and provides for a satisfactory immunity to noise and a video recorder. It also guarantees compatibility with the "quick search mode".

This process is particularly effective if a reduction in the speed of a video signal having a non-interlaced video format is utilized. If the available signal is interlaced, its format is converted before encoding,
This will generate a video signal that is not interlaced. Therefore, as shown in FIG. 17, one image is composed in frame units, and the first dimension is taken in the horizontal direction, the second dimension is taken in the vertical direction, and the third dimension is taken in the time direction. A three-dimensional block is formed, and a redundant component of a video signal is reduced by performing orthogonal transformation.

However, an actual television screen employs interlaced scanning as shown in FIG. This is an advantageous method for preventing flicker without increasing the amount of transmission information for transmitting moving image information. Accordingly, scanning of one screen is completed with half the number of scanning lines in FIG. In the next screen, the line that has not been scanned in the immediately preceding screen is scanned, thereby suppressing a decrease in the vertical resolution of the image. By this interlaced scanning, the number of screens transmitted in the same time becomes twice that in the case of sequential scanning, so that the occurrence of flicker is suppressed. This roughly scanned screen is called a field, and as shown in FIG. 19, one continuous frame is composed of two fields. In the NTSC (National Television System Committee) system, this is about 60 fields per second.

[0013]

In the conventional encoding method, since a three-dimensional block is formed from video signals that are not interlaced, the redundancy of the video signal is not necessarily effectively reduced for the interlaced video signal. Could not reduce. In particular, if an interlaced video signal having a large motion is not interlaced and scanned, a two-dimensional image in which spatial displacement and temporal displacement are mixed is formed. Not effective.

When a digital video signal of the interlaced scanning method is encoded, the spatial displacement between adjacent fields is converted into a temporal displacement due to the effect of the interlaced scanning even for a completely still image. A pseudo video component appears. Therefore, if weighting quantization is performed on higher-order transform coefficients after orthogonal transformation performed in the time direction for the purpose of reducing the amount of information at the time of a moving image, weighting quantization is also performed on the pseudo moving image component. Therefore, there is a problem that the image quality of the still image is deteriorated on the decoding side. In order to solve such a problem, it is necessary to determine whether the image is a moving image or a still image in units of three-dimensional blocks, and perform different levels of weighted quantization according to the moving image and the still image.

The present invention has been made in view of such circumstances, and an object of the present invention is to provide an encoding method and an encoding method capable of reducing the amount of information at the time of a moving image without discriminating between a moving image and a still image. An apparatus and an encoding / decoding apparatus are provided.

Another object of the present invention is to provide an encoding method, an encoding device, and an encoding / decoding device capable of reducing the amount of information in a moving image without deteriorating the image quality of a still image on the decoding side. It is in.

Still another object of the present invention is to provide an encoding method capable of more effectively reducing the redundancy of an interlaced video signal by matching the spatial positions of pixels in odd and even fields. An object of the present invention is to provide an encoding device and an encoding / decoding device.

[0018]

An encoding method, an encoding device, and an encoding / decoding device according to the first, second, sixth, and eighth aspects of the present invention provide a transform coefficient in which a pseudo moving image component does not appear due to the effect of interlaced scanning. Is characterized in that after weighting , quantization is performed, and quantization is directly performed on a transform coefficient in which a pseudo moving image component appears.

In the coding method according to the third and fourth aspects of the present invention, the transform coefficients in which the pseudo moving image component does not appear are weighted at a lower rate than the transform coefficients in which the pseudo moving image component appears due to the effect of the interlaced scanning. It is characterized in that coarse quantization is performed.

In the encoding method and the encoding apparatus according to the fifth and seventh aspects, the inter-pixel operation is performed on the interlaced digital video signal to determine the positions of the pixels in the vertical direction in the odd and even fields. After the adjustment, an operation such as orthogonal transformation and filtering is performed on pixels in a plurality of fields.

[0021]

[Action] first and second, the 6,8 invention, since the originally value weighted transform coefficients are zero is performed when a still image, information is not at all lost, image quality for still images in decoding side No degradation. On the other hand, in the case of a moving image, since the conversion coefficient is converted into zero or a small number by weighting, the information amount is reduced.

According to the third and fourth aspects of the present invention, in the case of a still image, the low-rate weighting is applied to the transform coefficient which is originally zero, so that the degree of compression of the information of the still image is small as a whole. . On the other hand, in the case of a moving image, the weight is converted to zero or a small number by this low-rate weighting, so that the amount of information is greatly reduced.

Further, in the fifth and seventh aspects, the two-dimensional spatial positions of pixels in the odd and even fields of the interlaced video signal are the same. Therefore, the signal configuration is accurate in the time direction, and large information compression is achieved by pixel calculation between fields.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be specifically described with reference to the drawings showing the embodiments.

(First Embodiment) FIG. 1 shows a configuration of an encoding / decoding device according to a first embodiment.
, 1 is an NTSC color television signal
The input terminal for inputting to SC decoder 2 is shown. The NTSC decoder 2 converts the input color television signal into a luminance signal (Y signal) and color difference signals (RY signal, BY signal).
And outputs it to an analog / digital (hereinafter referred to as A / D) converter 3. The A / D converter 3 converts the input signals into digital signals and outputs these digital signals to the field memory 4. Field memory 4
Is used to form a three-dimensional block by bundling two-dimensional blocks in a plurality of fields that are close to each other in the time direction, and converts the data in units of the three-dimensional block into a three-dimensional orthogonal transformation circuit 5.
Output to The three-dimensional orthogonal transform circuit 5 applies, for example, a three-dimensional DCT (Discrete Cosine Tr
ansform: discrete cosine transform) to obtain a transform coefficient, and output the obtained transform coefficient to the weighting quantizer 6. The weighting quantizer 6 weighs the obtained transform coefficients, quantizes them, and outputs the quantized data to a variable length encoder 7.
Output to The variable-length encoder 7 performs variable-length encoding on the input data using, for example, a Huffman code so that a short code is assigned to a frequently output. The encoded data is output via the output terminal 8. Above 1
A code system is constituted by the members No. to No. 8.

Reference numerals 11 to 18 denote constituent members of a decoding system, and reference numeral 11 denotes an input terminal for inputting encoded data to the variable length decoder 12. The variable-length decoder 12 returns the variable-length encoded data to the original quantized data, and outputs this data to the weighted inverse quantizer 13. The weighted inverse quantizer 13 returns the quantized data to the original data of the transform coefficient, and outputs the data to the three-dimensional inverse orthogonal transform circuit 14. The three-dimensional inverse orthogonal transform circuit 14 performs the three-dimensional inverse DCT to obtain the original three-dimensional block, and outputs this to the field memory 15. The field memory 15 returns the three-dimensional block to the original field screen, and converts digital luminance signals (Y signals) and color difference signals (RY signals, BY signals) into digital / analog (hereinafter, referred to as D / A). ) Output to the converter 16. The D / A converter 16 converts the input signal into an analog signal and outputs the analog signal to the NTSC encoder 17. The NTSC encoder 17 reproduces an NTSC color television signal from a luminance signal and a color difference signal.
The reproduced color television signal is output via an output terminal 18.

Next, the operation will be described. Generally, in order to compress image information, it is convenient to handle a luminance signal and a chrominance signal independently. Therefore, the input from the input terminal 1
The NTSC color television signal is converted into a luminance signal (Y signal) and a color difference signal (RY signal, BY signal) by the NTSC decoder 2.
After that, the signal is converted into a digital signal by the A / D converter 3. At this time, the sampling frequency is 1 for the Y signal.
3.5 MHz, the RY signal and the BY signal are 6.75 MHz.
Therefore, in the case of an NTSC color television signal, the number of effective samples of one horizontal line is 720 for the Y signal, RY signal,
Each of the BY signals becomes 360, and one field is composed of 262.5 horizontal lines. Of these, effective fields, for example, 250 horizontal lines are taken as one field unit, and eight fields are taken into the field memory 4. Then, while the data of the next eight fields is being taken in, the data of the three-dimensional block as shown in FIG. 2 is output from the field memory 4 to the three-dimensional orthogonal transformation circuit 5.

In FIG. 2, a three-dimensional block is configured with the horizontal direction being the primary direction, the vertical direction being the secondary direction, and the field direction (time direction) being the tertiary direction. Specifically, a 2 × 2 × 8 block is shown. Note that FIG.
As shown in the figure, in the odd field and the even field, for example, the (2i-1) th field and the 2ith field, the spatial positions of the pixels do not coincide in the vertical direction.
The upper left pixel of the 2i-th field is located 1/2 line below the upper left pixel of the (2i-1) -th field. The data transmitted in such a three-dimensional block unit is subjected to a three-dimensional DCT by a three-dimensional orthogonal transformation circuit 5. Then, the obtained transform coefficient is assigned to the weighted quantizer 6.
The weighted data is quantized, and the quantized data is encoded by a variable length encoder 7 using a Huffman code or the like, and output from an output terminal 8.

Here, it will be shown that information can be compressed by the above-described method. An example in which three-dimensional DCT of 8 × 8 × 8 pixels is performed on a part of data of one frame of a certain natural moving image will be described. The data obtained by uniformly quantizing the luminance signals of the natural moving image from the fields t = 0 to t = 7 to 8 bits are as shown in Table 1.

[0030]

[Table 1]

Next, a three-dimensional DCT is performed on this data, and the fractional part is rounded off.
The conversion coefficient as shown in (b) was obtained.

[0032]

[Table 2]

[0033]

[Table 3]

When the obtained transform coefficients are coded using a coded bitmap as shown in FIG. 3 and two-dimensional scanning as shown in FIG. 4, the code length of the three-dimensional data is 1902.
Bit. In FIG. 3, the horizontal axis represents the level of the conversion coefficient, and the vertical axis represents the number followed by zero, and the number in the figure represents the code length. FIG. 4 is a scan diagram often used in the case of two-dimensional DCT. In this embodiment, such two-dimensional scanning is repeated eight times.

On the other hand, the same data is converted and encoded into a format in which interlaced scanning is not performed as in the prior art. Table 3 shows the original data in a format not interlaced.

[0036]

[Table 4]

In this case, there are two three-dimensional blocks of 8 × 8 × 4 pixels. The conversion coefficients obtained by performing three-dimensional DCT on these two blocks are shown in Tables 4 (a) and 4 (b).

[0038]

[Table 5]

[0039]

[Table 6]

When these are coded using the coded bit map as shown in FIG. 3 and the two-dimensional scanning as shown in FIG. 4, the code length in each block is 1188 bits,
That is 1137 bits, for a total of 2325 bits.

As described above, in the present embodiment, it was possible to achieve about 16% information compression as compared with the conventional example. In this embodiment, encoding is performed on the four-frame image of the natural moving image and converted into a rate per second, which is about 25.3 Mbps. Similarly, when converted to the rate per second in the conventional example, it is about 30.2 Mb
ps. From this, it can be seen that information compression can be achieved not only for a part of the image but also for the entire image.

FIG. 5 shows the power distribution of the transform coefficients by the DCT in the third order (time direction) in the first embodiment. FIG. 5A shows a case of a moving image, and it can be seen that each of the conversion coefficients has power and a large amount of information. FIG. 5B shows a case of a still image. Originally, there should be no information change in the time direction in a still image, but since the spatial displacement is converted to the temporal displacement by interlaced scanning, when the DC component is set to the zeroth, the power is converted to the odd-numbered conversion coefficient. appear. This is related to the order N of the base vector of the DCT. FIG. 6 shows the DCT when N = 16.
Shows the basis vector of.

Here, the reason why the power appears only in the odd-numbered transform coefficients will be described based on the DCT definition equation. N
The point DCT is defined by the following equation.

[0044]

(Equation 1)

Here, considering the odd and even fields separately, y (i) is expressed by the following equation.

[0046]

(Equation 2)

In the case of a still image, since the image signals are the same between odd fields and between even fields, x (0) = x (2) =,..., = X (N−2). , X (1) = x (3) =,..., = X (N−1). Using this, y (i) is further expressed by the following equation.

[0048]

(Equation 3)

In the cosine function, cos α = −cos (π−
Since α) = − cos (π + α) = cos (2π−α), the following equations hold when i = 2, 4, 6,..., N−2.

[0050]

(Equation 4)

As described above, the even-numbered conversion coefficients excluding the DC component become zero, and power appears in the odd-numbered conversion coefficients. The weighting quantizer 6 weights only the even-numbered transform coefficients having no information of the still image,
Equivalently roughly quantize. FIG. 7A shows an example of a weighting coefficient for this weighting. In the case of a still image, this weighting is performed on a transform coefficient whose value is originally zero, and thus has no effect.

In the case of a moving image, the conversion coefficient is converted to zero or a small number by this weighting.
The amount of information is reduced by the operation of the subsequent variable length encoder 7 using Huffman coding or the like. The data that has been subjected to the variable length encoding by the variable length encoder 7 is output from an output terminal 8.

On the other hand, in the decoding system from the input terminal 11 to the output terminal 18, a process completely opposite to the above-described coding system from the input terminal 1 to the output terminal 8 is performed, and the original NTSC color television signal is obtained. Output from the output terminal 18. FIG. 7B shows an example of a weighting coefficient used at the time of decoding.

FIGS. 8 and 9 show other examples of weighting coefficients used in encoding and decoding. In this example, even-numbered conversion coefficients are weighted more coarsely than odd-numbered conversion coefficients. Therefore, it is possible to greatly reduce the amount of information at the time of a moving image while minimizing deterioration of the image quality of a still image. In the present embodiment, an example using three-dimensional orthogonal transform has been described. However, if orthogonal transform is performed in the time direction, it goes without saying that a method of processing a two-dimensional space is arbitrary.

As described above, in the first embodiment, the weighted quantization is performed only on the even-numbered transform coefficients in which the pseudo moving image component does not appear, or the pseudo moving image component is converted into the transform coefficient in which the pseudo moving image component does not appear. Compared to the odd-numbered transform coefficients that appear, they are weighted at a lower rate and quantized more coarsely, so without discriminating between moving images and still images,
Further, the amount of information at the time of moving images can be significantly reduced without deteriorating the image quality of still images.

(Second Embodiment) By the way, when a three-dimensional block is formed from the signal format of the interlaced scanning method, the spatial positions of the pixels in the odd and even fields in the vertical direction do not match. The possibility remains that redundancy can be reduced. In the second embodiment of the present invention, the interlaced digital video signal is subjected to an inter-pixel operation in each field in the vertical direction to adjust the spatial positions of the pixels in the odd and even fields, and then to perform three-dimensional orthogonalization. Conversion is performed.

In FIG. 10 showing the configuration of the encoding / decoding device according to the second embodiment, the same reference numerals as in FIG. 1 denote the same members. Numeral 9 in the code system is an encoder that quantizes and encodes the transform coefficients from the three-dimensional orthogonal transform circuit 5, and 10 is a digital luminance signal and color difference signal output from the A / D converter 3. This is a vertical interpolation filter that performs an inter-pixel operation in the vertical direction in each field to adjust the spatial position of pixels in odd and even fields. The vertical interpolation filter 10 outputs the data after the positioning to the field memory 4. Also, in the decoding system
19 is a decoder for restoring the encoded data to the original three-dimensional data format and outputting it to the three-dimensional inverse orthogonal transform circuit 14;
Is a vertical interpolation filter for restoring the original digital pixel data from the data in which the pixels output from the field memory 15 are aligned. The vertical interpolation filter 20
The restored pixel data is output to the D / A converter 16.

Next, the operation will be described. Pixel alignment and vertical interpolation filter in vertical interpolation filter 10
Except for the restoration of the original pixel data at 20, the operation of this embodiment is basically the same as that of the first embodiment described above.
Here, different points will be mainly described. FIG. 11 is a conceptual diagram showing pixels for explaining the operation in the second embodiment.

The luminance signal and the chrominance signal converted into digital signals by the A / D converter 3 are composed of an odd field and an even field as shown in FIG.
In the 1) field and the 2i-th field, the spatial positions of the pixels (indicated by the circles) do not match, and the pixels in the even fields are located ラ イ ン line below the pixels in the odd fields.

In order to align the pixels in the vertical interpolation filter 10, an oversampling technique is used. Now, in order to double the number of pixels, zero data is inserted every other in the vertical direction. The □ marks in FIG. 11B indicate the inserted zero data. When this data is passed through an interpolation low-pass filter, interpolation data is obtained for the inserted zero data, and the number of pixels is doubled. As the interpolation filter, for example, FIG.
An odd tap filter having an impulse response shown in FIG. As shown in FIG. 11C, the obtained interpolation data is decimated so as to leave pixels having the same spatial position, and the decimated data is output to the field memory 4. In the field memory 4, a three-dimensional block for each of a plurality of pixels is configured with the horizontal direction being the primary direction, the vertical direction being the secondary direction, and the time direction being the tertiary direction. In the three-dimensional orthogonal transformation circuit 5, the three-dimensional block constructed is
A dimensional DCT is performed.

In the decoding system, the three-dimensional block data subjected to the three-dimensional inverse orthogonal transformation is converted into field data whose pixel spatial positions match in the field memory 15 (FIG. 11D).
Becomes Here, in order to reproduce the original pixel data, when zero data is inserted and passed through an interpolation filter as in the case of the coding system, data as shown in FIG. 11E is obtained. Further, if the data of □ is thinned out, a digital signal (FIG. 11 (f)) of the same interlaced scanning method as that of FIG.

FIGS. 14, 15 and 16 conceptually show the above-mentioned situation on the frequency axis. In each figure, the vertical axis represents the frequency in the vertical direction, and the horizontal axis represents the frequency in the time direction. Fig. 14
11A is a diagram at the time of interlaced scanning corresponding to FIG. 11A, and information is woven in an oblique direction by the effect of offset subsampling. FIG. 15 corresponds to a state where the number of pixels is doubled by double oversampling (FIG. 11B), and a vertical low-pass filter extracts vertical low-frequency components.

FIG. 16 corresponds to a state in which the number of pixels has returned to the original state by the 1/2 thinning (FIG. 11C), and the high frequency components folded in the vertical direction are separated in the time direction. Work advantageously for information compression. On the other hand, the decoding system follows the reverse path to the coding system.
(E) and (f) correspond to FIGS. 16, 15, and 14, respectively, and the original interlaced scanning digital signal is restored.

In the above example, the oversampling and the １ ／ thinning-out method are used for pixel alignment, but the alignment can be performed immediately using an all-pass filter. For example, a filter having a response indicated by a circle in FIG. 12 may be applied to an odd field, and a filter having a response indicated by a circle may be applied to an even field.

Next, it will be shown that the same processing can be performed with a filter having an even number of taps. As an example of an even-tap interpolation filter, there is one having an impulse response as shown in FIG. In the interlaced scanning signal (FIG. 11A), zero data is inserted in every other direction in the vertical direction (FIG. 11B) as in the case of the filter of the odd tap, and the signal passes through the interpolation filter. Double oversampling interpolation data is obtained. However, since the center of gravity of the filter of the even-numbered tap is located not on the pixel but in the middle of the adjacent pixel, the position of the pixel after the filtering is moved by １ ／ line. 1
By the / 2 thinning-out, pixels having the same spatial position are obtained as shown in FIG. The frequency spectrum at this time is not different from that of the filter with the odd tap.

The signal input to the interpolation filter at the time of decoding is as shown in FIG. 11 (d '). If zero data is inserted and filtered, the signal is shifted by 1/4 line as in the case of encoding.
11 (e) is obtained. If the offset sub-sampling is performed in the same manner as in the case of the odd-tap filter, the original interlaced scanning digital signal (FIG. 11F) is obtained.

Also, similarly to the case of using the filter of the odd-numbered tap, the case of using the filter of the even-numbered tap can also be achieved by applying the response indicated by ● in FIG. 13 to the odd field and the response indicated by ○ to the even field. Thus, it is possible to immediately perform pixel alignment in the odd and even fields.

As described above, in the second embodiment, the position of the pixels of the odd and even fields in the two-dimensional space is adjusted by the vertical interpolation filter in the field with respect to the video signal of the interlaced scanning method, Since a three-dimensional block is formed by bundling the odd and even fields in the direction, and the three-dimensional block is orthogonally transformed and coded, information compression can be remarkably performed on the interlaced scanning video signal. The second embodiment and the first embodiment may of course be combined.

In the above embodiment, the Y signal is set to 13.5 MHz,
Although the so-called 4: 2: 2 component signal in which the RY signal and the BY signal are sampled at 6.75 MHz has been described as an example, it goes without saying that the frequency at the time of sampling is not limited to this.

[0070]

As described above, in the encoding method, the encoding apparatus, and the encoding / decoding apparatus of the present invention, weighted quantization is performed only on the transform coefficient in which the pseudo moving image component does not appear, or the pseudo moving image component is For the conversion coefficient where does not appear,
Compared to the transform coefficient where the pseudo video component appears, weighting at a lower rate and quantizing more coarsely,
The amount of information at the time of a moving image can be significantly reduced without determining a still image and without deteriorating the image quality of the still image.

Further, in the encoding method and the encoding apparatus of the present invention, the positions of the pixels of the odd and even fields in the two-dimensional space are aligned with the video signal of the interlaced scanning method by a vertical interpolation filter in the field. Thereafter, odd and even fields are bundled in the time direction to form a three-dimensional block, which is orthogonally transformed and coded. Therefore, information compression can be significantly performed on a video signal of the interlaced scanning method. There is.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an apparatus configuration according to a first embodiment of the present invention.

FIG. 2 is a conceptual diagram showing a three-dimensional block.

FIG. 3 is a two-dimensional bitmap diagram used when encoding transform coefficients.

FIG. 4 is a scanning diagram of a two-dimensional conversion coefficient.

FIG. 5 is a power distribution diagram of transform coefficients after orthogonal transform.

FIG. 6 is a diagram showing base vectors of a DCT.

FIG. 7 is a diagram illustrating an example of a weighting coefficient.

FIG. 8 is a diagram showing another embodiment of a weighting coefficient.

FIG. 9 is a diagram showing still another example of the weighting coefficient.

FIG. 10 is a block diagram showing a device configuration according to a second embodiment of the present invention.

FIG. 11 is a conceptual diagram showing pixels for explaining the operation of the second embodiment.

FIG. 12 is a diagram illustrating an example of an impulse response of an odd tap interpolation filter.

FIG. 13 is a diagram illustrating an example of an impulse response of an even tap interpolation filter.

FIG. 14 is a conceptual diagram showing a spectrum distribution at the time of interlaced scanning.

FIG. 15 is a conceptual diagram showing a spectrum distribution at the time of oversampling.

FIG. 16 is a conceptual diagram showing a spectrum distribution at the time of pixel alignment.

FIG. 17 is a conceptual diagram showing a conventional three-dimensional block of a system in which interlaced scanning is not performed.

FIG. 18 is a diagram for explaining the principle of interlaced scanning of a television screen.

FIG. 19 is a conceptual diagram showing the relationship between fields and frames in the standard television system.

[Explanation of symbols]

 2 NTSC decoder 3 A / D converter 4 Field memory 5 3D orthogonal transform circuit 6 Weighted quantizer 7 Variable length encoder 9 Encoder 10 Vertical interpolation filter 12 Variable length decoder 13 Weighted inverse quantizer 14 3D inverse Orthogonal transformation circuit 15 Field memory 16 D / A converter 17 NTSC encoder 19 Decoder 20 Vertical interpolation filter

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Yoshinori Asamura 1 Baba Zujo, Nagaokakyo-shi, Kyoto, Japan Mitsubishi Electric Corporation Inside the Electronic Product Development Laboratory (72) Inventor Shun Ito 1 Baba Zhou, Nagaokakyo-shi, Kyoto Mitsubishi Inside the Electronic Products Development Laboratory, Electric Co., Ltd. (72) The inventor, Manabu Tsukamoto, 1 Baba Zoshosho, Nagaokakyo City, Kyoto Prefecture Inside the Electronic Products Development Laboratory, Mitsubishi Electric Corporation (56) References ) SPIE Vol. 66 (1975) "Efficient Transmission of Pictorial Information" p58-69, (REAL-TIME VIDEO COMP RESSION ALGORITHIM FOR FOR HADAMARD TRANSFORM PROCESSING).

Claims (8)

(57) [Claims] 1. An encoding method for encoding a digital video signal of an interlaced scanning method, wherein the digital video signal is divided into blocks at least in a time direction to form a block, and the formed block is subjected to an orthogonal transformation. A step of obtaining a transform coefficient and a step of quantizing and coding the obtained transform coefficient, and weighting the transform coefficient whose value becomes zero when the blocked signal indicates a still image. after the encoding method and performing quantization. 2. An encoding method for encoding a digital video signal of an interlaced scanning system, wherein a horizontal pixel of each field is a primary component and a vertical pixel of each field is a secondary component. Forming a three-dimensional block for each of a plurality of pixels by using pixels of a plurality of fields at substantially the same position in the three-dimensional space as a third-order component; And a step of quantizing and encoding the obtained transform coefficients, and weighting the transform coefficients whose values become zero when the three-dimensionally blocked signal indicates a still image. Thereafter , quantization is performed. 3. An encoding method for encoding a digital video signal of the interlaced scanning method, wherein the digital video signal is divided into blocks at least in a time direction to form a block, and the formed block is subjected to an orthogonal transformation. Obtaining a transform coefficient, and weighting, quantizing and encoding the obtained transform coefficient,
An encoding method characterized in that, when a blocked signal indicates a still image, a transform coefficient whose value is zero is weighted at a lower rate than a transform coefficient that does not become zero. 4. An encoding method for encoding a digital video signal of the interlaced scanning system, wherein a horizontal pixel of each field is a primary component and a vertical pixel of each field is a secondary component. Forming a three-dimensional block for each of a plurality of pixels by using pixels of a plurality of fields at substantially the same position in the three-dimensional space as a third-order component; And a step of weighting, quantizing, and encoding the obtained transform coefficient. The transform coefficient having a value of zero when the three-dimensionally blocked signal indicates a still image has a value of zero. An encoding method characterized in that weighting is performed at a lower rate than a transform coefficient that does not result in a conversion factor. 5. An encoding method for encoding a digital video signal of a plurality of continuous fields of an interlaced scanning method, wherein an arithmetic operation is performed on the digital video signal using pixels in a vertical direction in each field, and an odd number is calculated. field, a step of bringing the spatial position of the pixel in the even field, horizontal image of each field
Element as the first-order component, and the pixels in the vertical direction of each field are
A plurality of second-order components having substantially the same position in a two-dimensional space
Using the pixels of the field as the third order component, for each of a plurality of pixels
Constructing a three-dimensional block, and the constructed third order
Performs 3D orthogonal transformation in units of original blocks
The step of obtaining coefficients and the three-dimensional block is a still image
Weighted for the transform coefficient that is zero in the case
Encoding the obtained transform coefficients . 6. An encoding apparatus for encoding a digital video signal of a plurality of continuous fields of an interlaced scanning method, wherein adjacent fields are horizontally and vertically adjacent in each field.
The two-dimensional block is composed of
3D block by bundling blocks into multiple continuous fields
Means for constructing a block, and
And the third order with horizontal, vertical and field directions as dimensions
Means for performing transformation by performing orthogonal transformation and three-dimensional
If the conversion coefficient has a value of zero when the
And, after weighting, the encoding apparatus characterized in that it comprises a means for performing quantization. 7. An encoding apparatus for encoding a digital video signal of a plurality of continuous fields of an interlaced scanning method, wherein each of the digital video signals is
Perform calculations using pixels in the vertical direction within the
Spatial position of pixel in number field, even field
Means to align the pixels and the horizontal pixels in each field.
As the primary component, the vertical pixel of each field is
The next component is a plurality of files that are almost the same in the two-dimensional space.
Means for forming a three-dimensional block for each of a plurality of pixels by using the pixels of the field as a third-order component, means for performing a three- dimensional orthogonal transform on the formed three -dimensional block , and obtaining transform coefficients
After weighting the transform coefficients whose value is zero when the dimensional block is a still image, sign the obtained transform coefficients.
Encoding apparatus characterized by comprising a means for reduction. 8. Encoding an input interlaced digital video signal of a plurality of continuous fields ,
Decodes encoded data into a digital video signal
In coding and decoding apparatus for obtaining, in each field
Pixels that are adjacent in the horizontal and vertical directions are grouped
A two-dimensional block that constitutes a lock.
Means for constituting a field component bundled three-dimensional block, for the three-dimensional block composed, horizontal, vertical, fees
Means for performing a three-dimensional orthogonal transformation in which the three-dimensional block is a still image.
Are weighted for the transform coefficients that have a value of zero
After that, means for performing quantization and encoding the quantized transform coefficient
Means for converting the encoded data into a
Means for obtaining a number and means for dequantizing the obtained transform coefficient
And inverse orthogonal transformation of the inversely quantized change coefficient to obtain a third order
The means for obtaining the original block and the obtained three-dimensional block are combined.
Means for obtaining an original digital video signal by decoding .