JP2001268571A - Moving image re-encoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a moving image re-encoder with small memory capacity. SOLUTION: The moving image re-encoder 10 includes a decoding part 12 to receive moving image data encoded by an MPEG system as input and to decode the encoded moving image data by reducing it to data with the predetermined number of bits and a re-encoding part 14 to be connected with the decoding part 12 and to encode the decoded moving image data by the MPEG system by reducing it to data with the number of bits to be determined by relation with the predetermined number of bits.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture re-encoding apparatus for decoding coded moving picture data and then re-encoding the same, and more particularly to decoding and re-encoding of moving picture data with a small memory. The present invention relates to a moving picture re-encoding device capable of performing encoding.

[0002]

2. Description of the Related Art Digital video data distribution using the Internet or the like has become widespread in recent years. On the computer that receives the video data distribution, depending on its processing capability,
The transmission condition (transmission timing, transmission rate) or the image size is changed.

Conventionally, MPEG (Moving Picture Experts)
In order to change the transmission condition or the image size of the moving image data encoded by the Group) method, the moving image data is decoded, the image size is changed if necessary, and the encoding is executed again.

Referring to FIG. 8, a moving picture re-encoding device 210 for changing the transmission condition and moving image size of moving image data is described.
Receives a moving image data encoded by the MPEG system as an input and decodes the moving image data;
An image size changing unit 213 connected to the decoding unit 12 for changing the image size of the decoded moving image data, and a moving image data connected to the image size changing unit 213 and having the image size changed are re-transferred to the MPEG format. , And a re-encoding unit 214 for encoding.

[0005] As the decoding unit 12, Japanese Patent Laid-Open No. 2000-00
A moving image decoding device disclosed in Japanese Patent No. 4440 is used. The decoding unit 12 aims to reduce the storage capacity of the reference image memory that temporarily stores the reference image data. For this reason, the decoding unit 12 reduces the number of bits after performing the Hadamard transform process.

[0006]

However, the re-encoding unit 2
Reference numeral 14 denotes a normal MPEG encoding device. For this reason,
Both the decoding unit and the re-encoding unit require a reference image memory, and there is a problem that a large memory capacity is required when the moving image re-encoding device 210 as a whole is viewed.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problem, and an object of the present invention is to provide a moving picture re-encoding device having a small memory capacity.

[0008]

A moving picture re-encoding device according to an aspect of the present invention includes a decoding unit for decoding coded moving image data while reducing it to data having a predetermined number of bits. And a re-encoding unit, which is connected to the decoding unit and encodes the decoded moving image data while reducing the number of bits to be determined in relation to a predetermined number of bits.

According to the number of bits resulting from data reduction in the decoding unit, the number of bits resulting from data reduction in the re-encoding unit is determined. For this reason, it is not necessary for the re-encoding unit to secure an unnecessarily large number of data bits even though information is lost due to bit reduction in the decoding unit. Therefore, the memory capacity of both the decoding unit and the re-encoding unit can be reduced, and a moving picture re-encoding device with a small memory capacity can be provided.

A moving picture coding apparatus according to another aspect of the present invention is connected to a decoding section for decoding coded moving picture data while reducing it to data of a predetermined number of bits, and a decoding section. An image size changing unit for changing the image size of the moving image data, and connected to the image size changing unit, the decoded moving image data is converted into data having the number of bits determined in relation to a predetermined number of bits. And a re-encoding unit that performs encoding while reducing the number.

According to the number of bits resulting from data reduction in the decoding unit, the number of bits resulting from data reduction in the re-encoding unit is determined. For this reason, it is not necessary for the re-encoding unit to secure an unnecessarily large number of data bits even though information is lost due to bit reduction in the decoding unit. Therefore, the memory capacity of both the decoding unit and the re-encoding unit can be reduced, and a moving picture re-encoding device with a small memory capacity can be provided.

[0012] Preferably, the image size changing unit changes the image size of the moving image data in the frequency domain.

[0013] More preferably, the image resizing unit comprises:
The image size of the moving image data is changed in the frequency domain, and the result of the change is subjected to inverse orthogonal transform.

By changing the image size in the frequency domain, it is possible to change the image size by removing only the high-frequency components, and it is possible to re-encode the moving image data without any strange feeling to human eyes.

[0015]

[First Embodiment] Referring to FIG. 1, a moving picture re-encoding apparatus 1 according to a first embodiment of the present invention will be described.
0 is a decoding unit 12 which receives moving image data encoded by the MPEG system as an input and decodes the moving image data.
And a re-encoding unit 14 connected to the decoding unit 12 and re-encoding the decoded image data by the MPEG method.

Referring to FIG. 2, decoding section 12 receives a variable-length code of a transform coefficient, decodes the variable-length code, and is connected to variable-length decoding section 22, Transform coefficient (quantized DC
An inverse quantization unit 24 that inversely quantizes a T (Discrete Cosine Transform) coefficient and converts it into a DCT coefficient; and a DCT coefficient sequence that is connected to the inverse quantization unit 24 and generated by the inverse quantization unit 24. Inverse DCT unit 26 for converting to data
And an adder 28 connected to the inverse DCT unit 26 for adding the output of the inverse DCT unit 26 and the reference image data.

The decoding unit 12 is further connected to an adder 28, and is connected to a Hadamard transform unit 30 for performing a fourth-order Hadamard transform process and a quantization process on the output of the adder 28, and a Hadamard transform unit 30. A switch 39 that receives the reproduced image data of the I (Intra) picture or the P (Predictive) picture from the Hadamard transform unit 30 and alternately switches between two outputs to output the reproduced image data; Connected to reference image memories 38 and 40 for storing I-picture or P-picture reproduced image data obtained from the Hadamard transform unit 30 via the switch 39, and a reference image memory 38 A reverse Hadamard transform unit 34 for reading out the reproduced image data in units of macroblocks and performing inverse Hadamard transform; Is connected to the image memory 40, including from the reference image memory 40 reads the reproduced image data of a macro block unit, an inverse Hadamard transform unit 36 performs inverse Hadamard transform.

The decoding unit 12 is further connected to the inverse Hadamard transform units 34 and 36, and the inverse Hadamard transform unit 3
4 and 36, read and average the image data, respectively, and averaging unit 32 for generating macroblock-based reference image data used for bidirectional predictive encoding; averaging unit 32, inverse Hadamard transform units 34 and 36; The state is determined according to the type of data output from the inverse DCT section 26 and connected to the ground terminal 52. The averaging section 32, the inverse Hadamard transform sections 34 and 36, and the ground terminal 52
A switch 46 for supplying data given from the adder 28 to the adder 28, a Hadamard transform unit 30, and a reference image memory 38.
B (Bidirectional) which is connected to one of the reference image memory 40 and output from the Hadamard transform unit 30.
y predictive) The reproduced image data of the picture, the reproduced image data of the I picture or P picture output from the reference image memory 38, and the reproduced image data of the I picture or P picture output from the reference image memory 40 are the original image. A switch 44 for changing the state so that the output is performed in the same order as the reproduction order.

The decoding unit 12 is further connected to the switch 44, performs an inverse Hadamard transform process on the reproduced image data output from the switch 44, and outputs an inverse Hadamard transform unit 42 for outputting moving image data. 3
0, which is connected to the inverse Hadamard transform units 34 and 36 and the switches 39, 44 and 46, receives a control signal including a macroblock type, and based on the control signal, receives the Hadamard transform unit 30, the inverse Hadamard transform units 34 and 3
6, and a CPU (Central Processing Unit) 48 for controlling the switches 39, 44 and 46, for receiving the variable length code of the motion vector, decoding the variable length code, and controlling the cutout position of the reference image. And a variable-length decoding unit 50 that supplies a motion vector as a control signal to the reference image memories 38 and 40.

Referring to FIG. 3, re-encoding section 14 receives moving image data in which frames output from inverse Hadamard transform section 42 of decoding section 12 are arranged in display order, And a subtractor 64 connected to the screen rearranging unit 62 for obtaining a difference between the frame to be coded and the reference image frame. A DCT unit 66 that divides the difference between the target frame and the reference image frame into 8 × 8 pixel units, and performs a discrete cosine transform process on each block;
DCT connected to the DCT unit 66 and output from the DCT unit 66
Quantizing unit 6 that performs quantization to reduce the amount of information of coefficients
8 is included.

The re-encoding unit 14 further includes a quantization unit 68
And a variable-length encoding unit 70 that performs variable-length encoding of the quantized DCT coefficient and reduces the number of bits without reducing the amount of information. In order to output the encoded data under transmission conditions (transmission timing, transmission rate, etc.) assumed by the re-encoding unit 14, the buffer unit 72 temporarily stores the encoded data.

The re-encoding unit 14 further includes a quantization unit 68
And a switch 69 connected to the switch 69 for changing the state in accordance with the type of image data to be encoded, and an inverse quantization unit 74 for returning the quantized DCT coefficients to the original state.
And an inverse DCT which is connected to the inverse quantization unit 74 and performs inverse discrete cosine transform processing on the DCT coefficients to restore the original frame.
And a T unit 76.

The re-encoding unit 14 further receives an output of a switch 98, which will be described later, and, depending on whether or not to perform motion compensation, a switch 94 for selecting whether to output the output of the switch 98 or not. And an adder 78 connected to the inverse DCT unit 76 for adding the output of the inverse DCT unit 76 and the output of the switch 94; and an adder 78 connected to the adder 78 for subjecting the output of the adder 78 to Hadamard transform, A Hadamard transform unit 90 for reducing the number, connected to the Hadamard transform unit 90, receiving reference image data for an I picture or a P picture from the Hadamard transform unit 90,
A switch 96 for alternately switching between two outputs to output reference image data, and a reference image for an I picture or a P picture which is connected to a corresponding output of the switch 96 and obtained from the Hadamard transform unit 90 via the switch 96 And reference image memories 86 and 88 for storing data.

The re-encoding unit 14 is further connected to a reference image memory 86, reads out reference image data from the reference image memory 86 in macroblock units, performs inverse Hadamard transform, and restores the reduced bits. An inverse Hadamard transform unit 80 for reading out reference image data from the reference image memory 88 in units of macroblocks, performing an inverse Hadamard transform, and restoring the reduced bits, and an inverse Hadamard transform unit 80 , And an averaging unit 84 for reading out and averaging the reference image data from the inverse Hadamard transform units 80 and 82 in block units.

The re-encoding unit 14 is further connected to reference image memories 86 and 88, and the reference image memory 86
And 88, a portion most similar to the macroblock to be coded is searched from the reference image data stored in the other, and a motion compensation unit 9 for selecting a prediction method.
2, the inverse Hadamard transform units 80 and 82, and the averaging unit 8
4, and the motion compensation unit 92 connected to the ground terminal 100.
A state is determined in accordance with the prediction method output from the control unit, and a switch 98 that supplies a reference image frame to the subtractor 64 and the switch 94 is included.

Referring to FIG. 2, decoding section 12 operates as follows. The variable-length code of the transform coefficient among the data encoded by the MPEG method is sent to the variable-length decoding unit 22. The variable length code of the motion vector is sent to the variable length decoding unit 50 and decoded. The variable length decoding unit 50
As a control signal for controlling the cut-out position of the reference image, the variable length code of the motion vector is stored in the reference image memory 38.
And to the reference image memory 40.

The variable length decoding section 22 decodes a variable length code of a transform coefficient. The inverse quantization unit 24 converts the transform coefficients (quantized DCT coefficients) obtained from the variable length decoding unit 22
Are inversely quantized and converted into DCT coefficients. Inverse DCT unit 26
Is the inverse DCT coefficient generated by the inverse quantization unit 24,
Perform T conversion processing. In the adder 28, the inverse DCT unit 2
Reference image data corresponding to the macroblock type is added to the image data output from 6 to generate reproduced image data. The reference image data is sent to the adder 28 via the switch 46. However, when the image data output from the inverse DCT unit 26 is reproduced image data for the intra-frame prediction code, the switch 46 selects the ground terminal 52 and the reference image data is not provided to the adder 28.

The reproduced image data in macroblock units obtained as a result of the addition by the adder 28 is sent to the Hadamard transform unit 30. The Hadamard transform unit 30 performs a fourth-order Hadamard transform process and a quantization process. The reproduced image data is divided into 2 × 2 blocks. The pixel value (8 bits) of each pixel in each block is represented by O (0,0), O (1,0), O (0,
1), O (1,1), and the pixel values of each pixel of the 2 × 2 block after the Hadamard transform are A (0,0), A (1,0), A (0,1), A ( 1, 1), the Hadamard transform can be expressed as the following equations (1) to (4).

A (0,0) = [O (0,0) + O (1,0) + O (0,1) + O (1,1)] / 2 (1) A (1,0) ) = [O (0,0) + O (1,0) -O (0,1) -O (1,1)] / 2 (2) A (0,1) = [O (0,0 ) -O (1,0) + O (0,1) -O (1,1)] / 2… (3) A (1,1) = [O (0,0) -O (1,0) -O (0,1) + O (1,1)] / 2 ... (4) A (0,0), A (1,0), A (0,1) and A (1,1) are This is an 8-bit value. That is, the value is a total of 32 bits. The Hadamard transform unit 30 quantizes the four pixels after the Hadamard transform so that the total number of bits becomes half, that is, 16 bits. For example, A (0,0), A (1,0), A (0,1) and A (1,1)
Are quantized so as to be 8 bits, 4 bits, 4 bits, and 0 bits, respectively.

When the reproduced image data in macroblock units obtained by the Hadamard transform unit 30 is reproduced image data for an I picture or a P picture, the reproduced image data is transmitted via a switch 39 to a reference image memory 38 or 40. Reference image memory 3
The size of the image data stored in 8 or 40 is の of the conventional size. The switch 39 is controlled by the CPU 48.

The inverse Hadamard transform unit 34 performs an inverse quantization process and an inverse Hadamard transform process on the reproduced image data in units of macroblocks read from the reference image memory 38 to create reference image data. That is, inverse quantization is performed so that the pixel values after quantization in which the bit allocation is 8: 4: 4: 0 are all 8 bits. Thereafter, the inverse Hadamard transform unit 34 performs an inverse Hadamard transform process. The values after inverse quantization are A '(0,0) and A' (1,
0), A '(0,1) and A' (1,1), and the pixel values after the inverse Hadamard transform are O '(0,0), O' (1,0), O '(0,1 ), O '(1,1)
The inverse Hadamard transform is represented by the following equations (5) to (8).

O ′ (0,0) = [A ′ (0,0) + A ′ (0,1) + A ′ (1,0) + A ′ (1,1)] / 2 (5) O '(1,0) = [A' (0,0) + A '(0,1) -A' (1,0) -A '(1,1)] / 2 ... (6) O' ( 0,1) = [A '(0,0) -A' (0,1) + A '(1,0) -A' (1,1)] / 2 ... (7) O '(1,1 ) = [A ′ (0,0) −A ′ (0,1) −A ′ (1,0) + A ′ (1,1)] / 2 (8) The inverse Hadamard transform unit 36 also performs the inverse Hadamard In a manner similar to the conversion unit 34, the reproduction image data in macroblock units read from the reference image memory 40 is subjected to an inverse quantization process and an inverse Hadamard transform process to create reference image data.

The pixel value O ′ (x, y) after the inverse Hadamard transform is
It is different from the pixel value O (x, y) input to the Hadamard transform unit 30. That is, since the number of bits is reduced by the Hadamard transform unit 30, the image after the inverse Hadamard transform causes deterioration in image quality and accumulation of errors. But,
By reducing the bits of the high-frequency component in the Hadamard transform unit 30, visual effects can be reduced.

The averaging unit 32 includes an inverse Hadamard transform unit 80
And 82 are respectively averaged to generate reference image data in macroblock units used for bidirectional predictive coding.

When the reference image is read from the reference image memory 38 or 40, the cut-out position is controlled based on the motion vector from the variable length decoding unit 50.

When the data output from the inverse DCT unit 26 is prediction error data for the forward prediction code or prediction error data for the backward prediction code, the switch 46 sets the inverse Hadamard transform unit 34 or 3
The selection is switched so as to select the reference image data output from 6.

When the data output from the inverse DCT unit 26 is prediction error data for a bidirectional prediction code,
The switch 46 is switched so as to select the reference image data output from the averaging unit 32.

The switch 44 reproduces the reproduced picture data for the B picture sent from the Hadamard transform unit 30 to the switch 44 and the I picture stored in the reference picture memory 38.
The CPU 48 controls the playback image data for the picture or P picture and the playback image data for the I picture or P picture stored in the reference image memory 40 to be output in the same order as the order of the original images (display order). You.

The reproduced image data output from the switch 44 is subjected to the same inverse quantization and inverse Hadamard transformation by the inverse Hadamard transform unit 42, and then output.

Referring to FIG. 3, re-encoding section 14 operates as follows. The screen rearranging unit 62 rearranges the image data input from the decoding unit 12 in display order into an order suitable for encoding (stream order). That is, the I-pictures and the P-pictures are rearranged so as to be earlier than the B-pictures in each cycle. For example, B, B, I, B,
The image data in the display order of B, P is I, B, B, P, B,
Sorted in the order of B.

The subtractor 64 calculates the difference between the macroblock of the image data received from the screen rearranging section 62 and the macroblock of the reference image data output from the switch 98. The DCT unit 66 receives the macroblock from the subtractor 64 and performs a discrete cosine transform in units of 8 × 8 pixels.
The quantization unit 68 receives the DCT coefficient from the DCT unit 66,
Quantize.

The variable length coding unit 70 has a quantized DC
The T coefficient is variable-length coded and stored in the buffer unit 72.
The encoded data stored in the buffer unit 72 is read so as to satisfy predetermined conditions such as a transmission timing and a transmission rate, so that MPEG encoded data is output.

On the other hand, when the image data to be encoded is an I picture or a P picture, the image data is used for predictive encoding of other image data. Therefore, the switch 69 is switched so as to connect the quantization unit 68 and the inverse quantization unit 74. The image data to be encoded is later
It will be stored in the reference image memory 86 or 88.

When the image data to be encoded is a B picture, the image data is not used for predictive encoding of other image data. Therefore, the quantization unit 68 and the inverse quantization unit 74 are not connected.

When the image data to be encoded is an I picture or a P picture, the inverse quantization unit 74 performs a process of restoring the inversely quantized DCT coefficients output from the quantization unit 68. The inverse DCT unit 76 performs an inverse discrete cosine transform process on the DCT coefficients output from the inverse quantization unit 74 to restore the original frame.

Switch 94 is connected to receive an output from switch 98 when performing motion compensation inter-frame prediction, and is connected to a ground terminal when motion compensation inter-frame prediction is not performed.

The adder 78 outputs the output of the inverse DCT unit 76,
The output of the switch 94 is added. Hadamard transform unit 9
0 performs Hadamard transform on the output of the adder 78 and performs processing to reduce the number of bits. The Hadamard transform is as shown in the above equations (1) to (4). The values after the Hadamard transform are 8-bit values. A (0,0), A (1,0), A (0,
Assuming that 1) and A (1,1), the Hadamard transform unit 90 quantizes the respective values to be 8 bits, 4 bits, 4 bits, and 0 bits. That is, the bit reduction is performed according to the same rule as the bit reduction in the Hadamard transform unit 30 of the decoding unit 12. The quantization halves the total number of bits.

If the image data in macroblock units obtained by the Hadamard transform unit 90 is for an I-picture or a P-picture, the image data is transferred via a switch 96 to a reference image memory 86 or 88.
Is stored in The size of the image data stored in the reference image memory 86 or 88 is の that of the conventional image data.

The inverse Hadamard transform unit 80 performs an inverse quantization process and an inverse Hadamard transform process on the image data in macroblock units read from the reference image memory 86 to create reference image data. That is, inverse quantization is performed so that the pixel values after quantization in which the bit allocation is 8: 4: 4: 0 are all 8 bits. Thereafter, the inverse Hadamard transform unit 80 performs the inverse Hadamard transform unit 34 described above.
Similarly, inverse Hadamard transform is performed according to equations (5) to (8).

Similarly to the inverse Hadamard transform unit 80, the inverse Hadamard transform unit 82 performs an inverse quantization process and an inverse Hadamard transform process on the macroblock image data read from the reference image memory 88. , Create reference image data.

The averaging unit 84 creates reference image data in macroblock units used for bidirectional predictive coding.

The motion compensation unit 92 searches the reference image memory 86 or 88 for a portion most similar to the macroblock to be coded, and performs forward prediction,
It decides which inter prediction between backward prediction and bidirectional prediction is to be performed. Alternatively, it is determined not to perform the inter-screen prediction.

The motion compensator 92 controls the switch 98 according to the type of inter prediction. That is, in the case of forward prediction or backward prediction, the switch 98 changes the state so as to supply the output of the inverse Hadamard transform unit 80 or 82 to the subtractor 64.

In the case of bidirectional prediction, the switch 98
The state is changed so that the output of the averaging unit 84 is supplied to the subtractor 64. When inter-picture prediction is not performed, the switch 98 is connected to the ground terminal 100 so that nothing is supplied to the subtractor 64.

The state of the switch 98 is classified according to the type of the macroblock to be processed as follows. When the macroblock is an I picture, inter-picture prediction is not performed, and thus the switch 98 is connected to the ground terminal 100.

When the macroblock is a P picture,
The switch 98 is connected to one of the inverse Hadamard transform unit 80, the inverse Hadamard transform unit 82, and the ground terminal 100 because forward prediction or inter-screen prediction is not performed.

When the macro block is a B picture,
The switch 98 is connected to any one of the inverse Hadamard transform unit 80, the inverse Hadamard transform unit 82, the averaging unit 84, and the ground terminal 100 because any of the above four states can be taken.

The switch 94 is turned on only when the image data to be encoded is a P picture, and the output of the switch 98 is supplied to the adder 78.

As described above, the state of each switch is determined according to the type of image data (macroblock) to be encoded or the search result in the motion compensator 92. Thus, the image data input from the re-encoding unit 14 is encoded.

At the time of quantization, the Hadamard transform unit 30 of the decoding unit 12 performs bit reduction to remove high-frequency components, thereby reducing the storage capacity of the reference image memories 38 and 40. For this reason, it is meaningless for the re-encoding unit 14 to hold high-frequency component data. Therefore, the Hadamard transform unit 90 of the re-encoding unit 14 performs bit reduction under the same conditions as the Hadamard transform unit 30. Therefore, the storage capacity of the reference image memories 86 and 88 can be reduced while maintaining the same image quality as the image data decoded by the decoding unit 12.

[Embodiment 2] Referring to FIG. 4, a moving picture re-encoding device 110 according to Embodiment 2 of the present invention
A decoding unit 112 that receives moving image data encoded by the PEG method as input and decodes and simultaneously changes the image size; and a decoding unit 112 connected to the decoding unit 112.
The image size changing unit 113 receives the image data in the frequency domain output from the image processing unit and changes the image size of the image data in the frequency domain. The image in the frequency domain is connected to the image size changing unit 113 and the image size is changed. And a re-encoding unit 114 for receiving the data and re-encoding the image data according to the MPEG method.

Referring to FIG. 5, decoding section 112 performs the operation shown in FIG.
In the configuration of the decoding unit 12 shown in FIG. 7, the inverse Hadamard transform unit 42 is omitted. That is, the decoding unit 11
2 outputs the decoded image data in the frequency domain. Other components are the same as those in FIG. 2, and the operation of the decoding unit 112 is also the same. Therefore, the detailed description will not be repeated here.

Referring to FIG. 6, re-encoding section 114 is different from re-encoding section 14 in FIG.
This is a configuration in which the inverse Hadamard transform unit 42 is provided between 2 and the subtractor 64. Image data in the frequency domain is input to the re-encoding unit 114. Therefore, it is necessary to return to the original image data once, and the inverse Hadamard transform unit 42 is provided in the re-encoding unit 114. Inverse Hadamard transform unit 4
2 performs the same operation as the inverse Hadamard transform unit 42 of the re-encoding unit 14. Therefore, the detailed description will not be repeated here.

The image size changing section 113 in FIG. 4 changes the image size in the frequency domain. For example, 3
Consider a case where image data of 20 × 480 pixels is converted into image data of 320 × 240 pixels. As a method of changing the size, a pixel value of the image data after the size change is obtained from two vertically adjacent pixels. For example, the Hadamard transform unit 30 of the decoding unit 112 converts the pixel values A (0,0), A (1,0), A (0,1), A (1,1) of the block of 2 × 2 pixels. It is assumed that the bit numbers of 8 bits, 4 bits, 0 bits, and 0 bits are allocated. In such a case, the vertical frequency components A (0,1) and A (1,1) are deleted, and A (0,0) and A (1,0) are replaced with a new 2 × 1 pixel block. Pixel value.

The Hadamard transform unit 90 of the re-encoding unit 114
Are corresponding to the number of bits of the pixel value output from the image size changing unit 113, and the quantized values A ′ (0,0), A ′ (1,0), A ′
(0,1) and A '(1,1) are 8 bits each,
The quantization is performed so that the number of bits becomes 8 bits or 4 bits. By performing such bit allocation, it is possible to reduce the storage capacity of the reference image memories 86 and 88 while preventing the image quality from being degraded in the re-encoding unit 114.

[Embodiment 3] Referring to FIG. 7, a moving picture re-encoding apparatus 120 according to Embodiment 3 of the present invention
A decoding unit 112 that receives moving image data encoded by the PEG method as input and decodes and simultaneously changes the image size, and is connected to the decoding unit 112, and changes the image size in the frequency domain and inverse Hadamard transform processing Are performed simultaneously, and a re-encoding unit 14 connected to the image size-change inverse Hadamard transform unit 123 and re-encoding the decoded image data by the MPEG method is included.

Decoding section 112 has the same configuration as decoding section 112 according to the second embodiment described with reference to FIG. Therefore, the detailed description will not be repeated here.

Re-encoding section 14 has the same configuration as re-encoding section 14 according to Embodiment 1 described with reference to FIG. Therefore, the detailed description will not be repeated here.

Image size change inverse Hadamard transform unit 123
Performs inverse Hadamard transform and image size change simultaneously according to equations (9) and (10). Here, it is assumed that the size of the image is halved in the vertical direction.

N (0,0) = A (0,0) + A (0,1) (9) N (0,1) = A (0,0) -A (0,1) (10 Here, A (u, w) indicates a pixel value after Hadamard transform of a block of 2 × 2 pixels. N (x, y) indicates the pixel value of the 2 × 1 pixel block after the inverse Hadamard transform.

When performing the inverse Hadamard transform, the vertical frequency components A (1,0) and A (1,1) are deleted. For this reason, the value after the inverse Hadamard transform is obtained according to the above equations (9) and (10).

In this embodiment, as in the second embodiment, the data output from the Hadamard transform unit 30 is used to reduce the storage capacity of the reference image memories 86 and 88 while preventing image quality deterioration. Is related to the number of bits of data output from the Hadamard transform unit 90.

The embodiments disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

[0074]

According to the present invention, a moving picture re-encoding apparatus having a small memory capacity can be provided.

Further, by changing the image size in the frequency domain, it is possible to change the image size by removing only the high-frequency components, and it is possible to re-encode the moving image data without any strangeness in human appearance.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a moving picture re-encoding device according to Embodiment 1 of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a decoding unit.

FIG. 3 is a block diagram illustrating a configuration of a re-encoding unit.

FIG. 4 is a block diagram showing a configuration of a video transcoder according to Embodiment 2.

FIG. 5 is a block diagram illustrating a configuration of a decoding unit.

FIG. 6 is a block diagram illustrating a configuration of a re-encoding unit.

FIG. 7 is a block diagram illustrating a configuration of a video transcoder according to Embodiment 3.

FIG. 8 is a block diagram illustrating a configuration of a conventional moving picture re-encoding device.

[Explanation of symbols]

10, 110, 120, 210 video re-encoding device,
12, 112 decoding unit, 14, 114, 214 re-encoding unit, 22, 50, 70 variable-length encoding unit, 24, 7
4 Inverse quantization unit, 26, 76 Inverse DCT unit, 28, 78
Adder, 30,90 Hadamard transform unit, 32,84
Averaging section, 34, 36, 42, 80, 82 inverse Hadamard transform section, 38, 40, 86, 88 reference image memory, 39, 44, 46, 69, 94, 96, 98 switch, 48 CPU, 52, 100 ground terminal, 62
Screen rearranging section, 64 subtractor, 66 DCT section, 68
Quantization unit, 72 buffer unit, 92 motion compensation unit, 11
3,213 image size change unit, 123 image size change inverse Hadamard transform unit.

Claims (10)

[Claims] A decoding unit configured to decode the encoded moving image data while reducing the encoded moving image data to data having a predetermined number of bits; and a decoding unit connected to the decoding unit and configured to decode the decoded moving image data. A re-encoding unit that performs encoding while reducing the number of bits to data determined in relation to a predetermined number of bits. 2. A decoding unit that decodes encoded moving image data while reducing the number of bits to a predetermined number of bits, and is connected to the decoding unit and changes an image size of the moving image data. An image size changing unit, connected to the image size changing unit, and re-encoding for reducing the number of bits determined in relation to the predetermined number of bits while decoding the decoded moving image data; And a moving image re-encoding device. 3. The decoding unit, comprising: a first adder for adding reproduced image data obtained from a variable length code of a transform coefficient and reference image data; and a decoder connected to the first adder, A first orthogonal transformation unit for orthogonally transforming the output of the adder of 1 and reducing the result of the orthogonal transformation to data having a predetermined number of bits; and a first orthogonal transformation unit connected to the first orthogonal transformation unit. A first reference image memory for storing an output of the orthogonal transformation unit, and a reference image data which is connected to the first reference image memory and generates reference image data from the data stored in the first reference image memory. Reference image data generating means for supplying the first orthogonal adder, the first orthogonal transform unit and the first reference image memory, and the output of the first orthogonal transform unit Stored in the first reference image memory A switch that receives data and selectively outputs one of them based on a control signal; and a first inverse orthogonal transform unit that is connected to the switch and that inversely orthogonally transforms data received via the switch. Including,
The moving picture re-encoding device according to claim 1. 4. A re-encoding unit, comprising: a screen rearranging unit for rearranging decoded moving image data in a predetermined order; and a re-encoding unit connected to the screen rearranging unit, the re-encoding unit comprising: A subtractor that takes a difference between the reference image data, a second orthogonal transform unit that is connected to the subtractor, and orthogonally transforms an output of the subtractor, and that is connected to the second orthogonal transform unit, A variable-length encoding unit that performs variable-length encoding on the output of the second orthogonal transformation unit; and a second inverse unit that is connected to the second orthogonal transformation unit and performs an inverse orthogonal transformation on the output of the second orthogonal transformation unit. An orthogonal transform unit; a second adder connected to the second inverse orthogonal transform unit, for adding an output of the second inverse orthogonal transform unit and reference image data; and an output of the second adder. Orthogonal transformation and the result of the orthogonal transformation in relation to the predetermined number of bits A third orthogonal transform unit that reduces the data to a predetermined number of bits; a second reference image memory that is connected to the third orthogonal transform unit and stores an output of the third orthogonal transform unit; Reference image data connected to a second reference image memory, for generating reference image data from data stored in the second reference image memory, and supplying the reference image data to the subtractor and the second adder The moving picture re-encoding device according to claim 3, further comprising a generation unit. 5. The moving image re-encoding device according to claim 2, wherein the image size changing unit changes the image size of the moving image data in a frequency domain. 6. The decoding unit, wherein: reproduced image data obtained from a variable length code of a transform coefficient;
A first adder for adding the reference image data; and a first adder connected to the first adder, for orthogonally transforming the output of the first adder, and converting the result of the orthogonal transform to a predetermined number of bits. A first orthogonal transformation unit for reducing data, a first reference image memory connected to the first orthogonal transformation unit, and storing an output of the first orthogonal transformation unit; and a first reference image. Image data generating means for generating reference image data from the data stored in the first reference image memory and supplying the reference image data to the first adder; Connected to a conversion unit and the first reference image memory, receives the output of the first orthogonal transformation unit and the data stored in the first reference image memory, and receives one of them based on a control signal. And a switch for selectively outputting The moving picture re-encoding device according to claim 5, comprising: 7. The re-encoding unit, further comprising: a screen rearranging unit that rearranges the decoded moving image data in a predetermined order; and a reordering unit connected to the rearranging unit, and inverts an output of the rearranging unit. A first inverse orthogonal transform unit for orthogonally transforming, a subtractor connected to the first inverse orthogonal transform unit for taking a difference between the image data to be encoded and the reference image data, and a subtractor connected to the subtractor. A second orthogonal transform unit for orthogonally transforming the output of the subtractor, and a variable length encoding unit connected to the second orthogonal transform unit for variable length encoding of the output of the second orthogonal transform unit. A second inverse orthogonal transform unit connected to the second orthogonal transform unit and inversely orthogonally transforming the output of the second orthogonal transform unit; and a second inverse orthogonal transform unit connected to the second inverse orthogonal transform unit. A second adder for adding the output of the inverse orthogonal transform unit 2 and the reference image data; A third orthogonal transform unit that orthogonally transforms an output of the second adder and reduces a result of the orthogonal transform to data having a number of bits determined in relation to the predetermined number of bits; A second reference image memory connected to the orthogonal transformation unit and storing the output of the third orthogonal transformation unit; and a second reference image memory connected to the second reference image memory and stored in the second reference image memory 7. The moving image re-encoding device according to claim 6, further comprising: reference image data generating means for generating reference image data from the obtained data and supplying the generated reference image data to the subtractor and the second adder. 8. The moving image re-encoding device according to claim 2, wherein the image size changing unit changes the image size of the moving image data in a frequency domain, and performs an inverse orthogonal transform on a result of the change. 9. The decoding unit according to claim 1, wherein: the reproduced image data obtained from a variable-length code of a transform coefficient;
A first adder for adding the reference image data; and a first adder connected to the first adder, for orthogonally transforming the output of the first adder, and converting the result of the orthogonal transform to a predetermined number of bits. A first orthogonal transformation unit for reducing data, a first reference image memory connected to the first orthogonal transformation unit, and storing an output of the first orthogonal transformation unit; and a first reference image. Image data generating means for generating reference image data from the data stored in the first reference image memory and supplying the reference image data to the first adder; Connected to a conversion unit and the first reference image memory, receives the output of the first orthogonal transformation unit and the data stored in the first reference image memory, and receives one of them based on a control signal. And a switch for selectively outputting The moving picture re-encoding device according to claim 8, comprising: 10. A re-encoding unit, comprising: a screen rearranging unit that rearranges decoded moving image data according to a predetermined order; and a re-encoding unit that is connected to the screen rearranging unit and encodes image data to be encoded. A subtractor that takes a difference between the reference image data, a second orthogonal transform unit that is connected to the subtractor, and orthogonally transforms an output of the subtractor, and that is connected to the second orthogonal transform unit, A variable-length encoding unit that performs variable-length encoding on the output of the second orthogonal transform unit; and a first inverse unit that is connected to the second orthogonal transform unit and performs an inverse orthogonal transform on the output of the second orthogonal transform unit. An orthogonal transform unit; a second adder connected to the first inverse orthogonal transform unit, for adding an output of the first inverse orthogonal transform unit and reference image data; and an output of the second adder. The orthogonal transform is performed, and the result of the orthogonal transform is related to the predetermined number of bits. A third orthogonal transform unit that reduces the data to the number of bits determined by: a second reference image memory that is connected to the third orthogonal transform unit and stores an output of the third orthogonal transform unit; A reference image connected to the second reference image memory, for generating reference image data from data stored in the second reference image memory, and supplying the reference image data to the subtractor and the second adder The moving picture re-encoding device according to claim 9, further comprising a data generation unit.