JPH07154788A - Encoder for video signal - Google Patents

Abstract

PURPOSE:To provide an encoder for a video signal which encodes a picture, where the amplitude of the luminance signal or the color signal is long or a complicated pattern is included, to prevent the distortion in the decoded picture from being conspicuous. CONSTITUTION:An inputted video signal passes the low band pass filter 10, and the value forecasted in accordance with already encoded video data is subtracted from this video signal by an operation means 30. An energy calculating means 20 calculates the energy of picture data to be encoded; and when the energy is large, the cut-off frequency of the filter means 10 is reduced. Encoded video data is used for calculation by a decoding means 40, a picture memory 70, a forecasting means 80, or the like and is encoded by a variable length code encoding means 45 and is transmitted or stored through a buffer memory 90 thereafter.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding device for digitally coding a video signal for transmission or storage.

[0002]

2. Description of the Related Art In recent years, along with the progress of information-oriented society, there is an increasing demand for transmitting moving images to others through time and distance barriers. In the era of full-fledged practical application of digital technology, it has become possible to record and reproduce video signals with a recording device and to transmit long distances using a communication network.
Since a digital video signal has an extremely large amount of information as compared with an audio signal, development of a high efficiency coding technique is indispensable for efficiently recording and transmitting this. Video signal coding methods are being standardized on a global scale, and are expected to be widely applied to various devices.
Conventionally, as a moving image encoding method, there is a method as disclosed in Japanese Patent Publication No. 4-139985. Hereinafter, an example of the above-described conventional video signal encoding apparatus will be described with reference to the drawings.

FIG. 4 shows a block diagram of a conventional video signal encoding apparatus. In FIG. 4, 1 is a video signal input terminal, 2 is an output terminal, 30 and 60 are arithmetic means, 40 is an encoding means, 45 is a variable length encoding means, 50 is a decoding means, 70 is an image memory, and 80. Is a prediction means, and 90 is a buffer memory.
FIG. 5 shows a block diagram of an example of the encoding means 40 in FIG. In FIG. 5, reference numeral 41 is an orthogonal transformation means, and 42 is a quantization means. Further, FIG. 6 shows zigzag scanning for image data. FIG. 7 is a graph showing an example of a graph showing the relationship between the transform coefficient (horizontal axis) input to the quantizer and the quantization result (vertical axis).

The operation of the video signal coding apparatus configured as described above will be described below with reference to FIGS. 4, 5, 6 and 7. The video signal given to the input terminal 1 is input to the encoding means 40 after the predicted value is subtracted from the video data already encoded by the arithmetic means 30 to reduce the variance of the signal to be actually encoded. To be done. Here, two-dimensional orthogonal transformation such as discrete cosine transformation, discrete wavelet transformation, and discrete Fourier transformation is performed on the block-shaped image data of 8 × 8 pixels by the orthogonal transformation means 41. By applying orthogonal transformation, the video signal is in the spatial domain
The (actual pixel value) is converted to the frequency domain (conversion coefficient). As shown in FIG. 6, conversion coefficients are obtained in the same number as the original number of pixels.

Next, the transform coefficient is input to the quantizing means 42 and quantized. FIG. 7 is a graph showing the relationship between the transform coefficient (horizontal axis) and the quantization result (vertical axis) input to the quantizing means shown in FIG. , When it is large, it becomes like a broken line. All the transform coefficients are quantized with a step width corresponding to each frequency.

The quantized result is input to the variable length coding means 45 of FIG. 4 and, at the same time, to the decoding means 50 for use in the next processing. Here, the reverse processing to that performed by the encoding means 40 is performed in just the reverse order. Decoding means 50
The result of (1) is added to the value predicted from the video data already encoded by the calculation means 60, and the same image as the result decoded by the decoding device is obtained (local decoding). This output is delayed by one video frame in the image memory 70 and used by the predicting means 80 for predicting the image of the next frame. The image prediction method is the previous frame itself (frame differential encoding)
Alternatively, the sum of products obtained by multiplying the values of the previous several frames by appropriate coefficient times is used.

The variable length coding means 45 performs zigzag scanning as shown in FIG. 6, performs variable length coding such as run length coding, and then stores it in the buffer memory 90. The reason why zigzag scanning is performed is that the coefficient in the high frequency band is generally small and there are many coefficients that become zero. Therefore, when scanning as shown in FIG. 6, continuous zero data, that is, zero run, can be efficiently captured. Also, as an example of run-length coding, two-dimensional run-length coding is used in which a table is drawn by a combination of the length of zero run and the coefficient value immediately following it. In this case, a short code is assigned to a frequently appearing pattern, for example, a coefficient having a short run length and a small absolute value, so that the generated code amount can be reduced as a whole.

The occupancy of the data in the buffer memory 90 of FIG. 4 is constantly monitored, and the quantization step width of the quantizing means 42 in the encoding means 40 is determined so that the occupying amount becomes substantially constant. Generally, when the buffer memory 90 is about to overflow, the quantization step width is increased. This is because, when the quantization step width becomes large, the coefficient having a particularly small amplitude in the high frequency band becomes zero among the conversion coefficients, and the subsequent variable length coding means 45 reduces the data generation amount.
When increasing the quantization step size, considering the human visual characteristics and not using the same step size for all coefficients, generally the DC component is relatively small and the high frequency component is Enlarge.

[0009]

However, the video signal coding apparatus having such a configuration, when coding an image in which the amplitude of a luminance signal or a chrominance signal is large, or an image including a complicated picture, is quantized. Since a large number of transform coefficients having a large absolute value that require a long generated code length are generated, it is necessary to increase the quantization step width for constant rate coding, which causes a problem that significant image distortion occurs. there were.

SUMMARY OF THE INVENTION The present invention solves the above-mentioned problems of the prior art, and an object of the present invention is to provide a video signal encoding device in which distortion in a decoded image is not noticeable.

[0011]

In order to achieve this object, a video signal coding apparatus of the present invention uses a video signal as an input, and in a video signal coding apparatus for coding, the video signal is filtered. Filter means for wave and energy calculating means for calculating the energy of the image data, and a filter means for filtering the video signal, and a transform coefficient for orthogonal transform coding of the image data to output it. An energy calculation means for calculating energy, and a difference signal between the predicted image based on the encoded output of the image data and the output of the filter means, and when the energy output by the energy calculation means is large, The cutoff frequency of the filter means is reduced.

[0012]

According to the present invention, when the image data to be coded or its transform coefficient has a large energy, the cutoff frequency of the filter means of the image input section is lowered to reduce the generated code amount. Since the encoding step width can be reduced, it is possible to perform encoding in which distortion is inconspicuous.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A video signal encoding apparatus according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a block diagram of a video signal coding apparatus according to a first embodiment of the present invention. The same effects as those of the conventional example are designated by the same reference numerals, and detailed description thereof will be omitted.
In FIG. 1, 10 is a filter means and 20 is an energy calculation means. FIG. 2 shows the frequency characteristic of the filter means 10. The operation of the video signal encoding apparatus configured as described above will be described below with reference to FIGS. 1 and 2.

The video signal supplied to the input terminal 1 is first subjected to low-frequency component extraction by the filter means 10 and then the arithmetic means 30.
At, the predicted value is subtracted from the already encoded video data, and the result is input to the encoding means 40. Here, the encoding means 40
Is similar to that shown in FIG. Coding means 40
The result of is input to the variable length coding means 45 and at the same time,
It is input to the decryption means 50 for use in the next process. Here, the reverse processing to that performed by the encoding means 40 is performed in just the reverse order. The result of the decoding means 50 is added to the value predicted from the video data already encoded by the calculation means 60 to obtain a locally decoded image. This output is delayed by one video frame in the image memory 70 and used by the predicting means 80 for predicting the image of the next frame.

On the other hand, the variable length coding means 45 performs the same processing as in the conventional example, and the output thereof is guided to the buffer memory 90. The occupied amount of data in the buffer memory 90 is constantly monitored, and the quantization step width of the quantizing means in the encoding means 40 is determined so that the occupied amount is almost constant. The energy calculation means 20 performs the following energy calculation as an example for the output of the calculation means 30, that is, for the image data of the image block actually used for encoding.

[0016]

## EQU1 ## m = Σx / N

[0017]

Σ ² = Σ (x−m) ² or σ ² = Σ (x−m) ² / N where Σ is the total sum for all pixels in the pixel block, and N is the total sum in the pixel block. The number of pixels, x is a pixel value.

The filter means 10 changes the cutoff frequency depending on the magnitude of σ ² . That is, as shown in FIG. 2, the cutoff frequency is changed, and when the energy is large, the cutoff frequency of the filter means 10 is lowered, and when not, it is increased. When the cutoff frequency of the filter means 10 is lowered, the high frequency coefficient generated when the image data is encoded is reduced, so that the code generation amount is reduced. As a result, the quantization step width can be reduced, and distortion during decoding is less noticeable. Further, the cutoff characteristic of the filter means 10 may be changed in the next frame after performing energy calculation for all pixel blocks in the frame, or may be changed immediately in the pixel block to be processed next.

Next, FIG. 3 shows a block diagram of a video signal encoding apparatus according to a second embodiment of the present invention. In FIG.
The same effects as those of the first embodiment are designated by the same reference numerals and detailed description thereof will be omitted. In the second embodiment, the energy calculation by the energy calculation means is performed using the conversion coefficient instead of the image data in the spatial domain. Since the DC component obtained as a result of the orthogonal transformation is an amount proportional to the average value m, it is multiplied by a constant to obtain m. σ ² is the same as (Equation 2), or the sum of absolute values of conversion coefficients is used. Similar to the first embodiment, by changing the cutoff characteristic of the filter means 10 depending on the magnitude of σ ² , the same effect as that of the first embodiment can be obtained.

As described above, according to the present embodiment, when the image data to be encoded itself or the energy of its transform coefficient is large, the cutoff frequency of the filter means of the image input section is lowered to reduce the generated code amount. As a result, the quantization step width can be reduced, and thus encoding with less noticeable distortion becomes possible.

Although the variable length coding means 45 performs run length coding without distinguishing between the DC component and the AC component in this embodiment, the DC component is different from the DC component of another block and the difference code. It is possible to achieve the same effect even when the AC component is run-length encoded. Furthermore, in the present embodiment, the case of interframe coding has been described, but intraframe coding or coding that adaptively switches between intraframe coding and interframe coding may be used, and subband coding technology may be applied to these. The same applies to a combination of motion compensation technology and the like.

Further, although the case of the frame unit coding is described in the present embodiment, the same applies to the pixel block or field unit coding. Further, although the case where the video signal is a moving image has been described in the present embodiment, the same effect can be obtained even when the image signal is a still image.

[0023]

As described above, according to this embodiment,
When the energy of the image data to be encoded itself or its transform coefficient is calculated and the energy is large, the cutoff frequency of the filter means of the image input section is lowered to reduce the generated code amount, so the quantization step width should be made small. Therefore, there is an effect that distortion in the decoded image can be made inconspicuous.

[Brief description of drawings]

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

FIG. 2 is an explanatory diagram showing frequency characteristics of the filter means in the embodiment of the present invention.

FIG. 3 is a block diagram showing a configuration of a video signal encoding device according to a second embodiment of the present invention.

FIG. 4 is a block diagram showing a configuration of a conventional video signal encoding device.

FIG. 5 is a block diagram showing a configuration of a coding unit in a conventional video signal coding apparatus.

FIG. 6 is an explanatory diagram of zigzag scanning.

FIG. 7 is an explanatory diagram of a quantization step width.

[Explanation of symbols]

1 ... input terminal, 2 ... output terminal, 10 ... filter means,
20 ... Energy calculation means, 30, 60 ... Calculation means, 40
... Encoding means, 41 ... Orthogonal transformation means, 42 ... Quantization means, 45 ... Variable length coding means, 50 ... Decoding means, 70
… Image memory, 80… Prediction means, 90… Buffer memory.

Claims (3)

[Claims] 1. A video signal encoding apparatus which receives a video signal as input and which encodes the video signal, comprising: filter means for filtering the video signal; and energy calculation means for calculating energy of image data. The video signal encoding device, wherein the filter means lowers the cutoff frequency when the energy output from the energy calculating means is large. 2. A video signal coding apparatus which receives a video signal as an input and which encodes the video signal, and a filter means for filtering the video signal and a transform coefficient output by orthogonal transform coding the image data. Energy calculating means for calculating the energy of the video signal, and the filter means lowers the cutoff frequency when the energy output by the energy calculating means is large. 3. The video signal coding apparatus according to claim 1, wherein the image data is a difference signal between the predicted image based on the coded output and the output of the filter means. .