JP3818484B2 - Decoding apparatus and recording medium for encoded moving image data - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an apparatus for decoding encoded moving image data, and more particularly to an apparatus for decoding encoded moving image data that can be used for a personal computer with low processing performance by reducing the load of moving image decoding processing.
[0002]
[Prior art]
FIG. 12 shows a conventional example of a decoding apparatus for a compressed and encoded moving image. In FIG. 12, encoded moving image data is input to a variable length decoder 11 and subjected to variable length decoding. The decoded quantized coefficient a, that is, the quantized discrete cosine transform coefficient, is input to the inverse quantizer 12, while the decoded motion vector information b is input to the motion compensated predictor 17. The quantization coefficient a is inversely quantized by the inverse quantizer 12, and the discrete cosine transform coefficient F (u, v) is input to the inverse discrete cosine transformer 31. The motion compensated predictor 17 extracts predicted image data used for prediction from the image stored in the frame memory 18 using the motion vector information b.
[0003]
The encoding mode information c decoded by the variable length decoder 11 controls the switch means 19. When the coding mode is the in-plane coding mode, the switch means 19 is turned off, and the output f (x, y) from the inverse discrete cosine transformer 13 is not added by the adder 16 and is output as the decoded image output r. It is output as (x, y) and stored in the frame memory 18.
[0004]
On the other hand, when the coding mode is other than the in-plane coding mode, the switch means 19 is turned on, and the output f (x, y) from the inverse discrete cosine transformer 31 is the motion compensated prediction image c (x, y) and The signals are added by the adder 16 and output as a decoded image output r (x, y) and are stored in the frame memory 18.
[0005]
In video decoding, inverse discrete cosine transform has the largest processing load. For example, BGLee, "A new algorithm to compute the discrete cosine transform," IEEE Trans.Acoust., Speech, and Signal Processing, vol.ASSP- 32, pp / 1243-1245, Dec. 1984. etc., a fast inverse discrete cosine transform algorithm is used.
[0006]
When higher speed is required, a method of reducing the decoding process by thinning the number of screens to be decoded is used. For example, a method is also used in which only an image that has been intra-coded (intra-coded) is decoded, and an image that has been encoded other than intra-coded is not decoded.
[0007]
[Problems to be solved by the invention]
However, for example, when performing decryption processing with software using a personal computer or the like, if the processing performance of the personal computer is low, there is a problem that the processing cannot be performed in time even with such high-speed processing, and the number of playback screens is greatly reduced. .
[0008]
An object of the present invention is to solve the above-mentioned problems of the prior art and to provide a decoding apparatus for encoded moving image data that can be processed at a higher speed than the conventional processing speed. Another object of the present invention is to provide an apparatus for decoding encoded moving image data that can greatly reduce the load of moving image decoding processing without causing a significant deterioration in image quality during reproduction and a reduction in the number of reproduced screens.
[0009]
[Means for Solving the Problems]
In order to achieve the above object, the present invention provides a discrete cosine transform coefficient that is dequantized smaller than the discrete cosine transform base on the encoding side. And does not include the basis of only the DC component Means for changing to a transform coefficient for the basis, and the transform coefficient changed by the means is smaller than the discrete cosine transform base on the encoding side. And does not include the basis of only the DC component A first feature is that it comprises means for performing inverse transform using inverse discrete cosine transform using a base, and means for transforming the image data subjected to inverse discrete cosine transform into the same size as that on the encoding side. .
[0010]
In addition, the present invention provides a computer-readable recording medium on which the above respective means (functions) are recorded. 2 There are features.
[0011]
Said first, first 2 According to the feature, the dequantized discrete cosine transform coefficient is smaller than the discrete cosine transform basis on the encoding side. Conversion factor for Since the inverse transform is performed using an inverse discrete cosine transform using a base smaller than the discrete cosine transform base on the encoding side, the processing speed required for the inverse transform can be improved, and the overall decoding processing speed is greatly increased. Can be improved.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the first exemplary embodiment of the present invention. In addition, the same code | symbol as FIG. 12 shows the same or equivalent. In this embodiment, the inverse discrete cosine transform is performed on the decoding side using a basis smaller than the discrete cosine transform on the encoding side.
[0014]
The encoded data or the encoded moving image data is input to the variable length decoder 11. The variable length decoder 11 decodes the quantized discrete cosine transform coefficient a, motion vector information b, coding mode information c, and the like. The decoded quantized discrete cosine transform coefficient a is input to the inverse quantizer 12, the motion vector information b is input to the motion compensated predictor 17, and the encoding mode information c controls the switch means 19 described later.
The quantized discrete cosine transform coefficient a input to the inverse quantizer 12 is inversely quantized to output a discrete cosine transform coefficient F (u, v). The discrete cosine transform coefficient F (u, v) is input to the scaler 13, and the coefficient data is scaled. As for the scaling method, each discrete cosine transform coefficient is changed along the equation given by equation (1) in FIG.
[0015]
Here, F (u, v) and F ′ (u, v) are a discrete cosine transform coefficient input to the scaler 13 and a discrete cosine transform coefficient after scaling processing, respectively. N × N (N is a positive even number) is the base size of the discrete cosine transform on the encoding side, u and v are the coordinates of the discrete cosine transform coefficients in the horizontal direction and the vertical direction, respectively, u = 0, 1, ..., (N / 2 ^p1 −1), v = 0, 1,..., (N / 2 ^p2 -1). p1 and p2 are parameters (integers) for determining the base size of the small basis inverse discrete cosine transform in the horizontal direction and the vertical direction, respectively. The base size is N / 2 in the horizontal and vertical directions, respectively. ^p1 , N / 2 ^p2 It is. For example, if p1 = p2 = 1, the base size is N / 2 × N / 2.
[0016]
In addition, the following scaling (2) can also be used.
F ′ (u, v) = F (u, v) / {(2 ^{p1 / 2} ) X (2 ^{p2 / 2} )}… (2)
The discrete cosine transform coefficient F ′ (u, v) after the scaling processing is input to the small basis inverse discrete cosine transformer 14. In the small basis inverse discrete cosine transformer 14, the basis size in the horizontal direction and the vertical direction is N / 2. ^p1 , N / 2 ^p2 Compared to the conventional case, the discrete cosine transform coefficient F ′ (u, v) is inversely transformed by a small basis inverse discrete cosine transform, and an image f ′ (i, j) is output.
[0017]
The image f ′ (i, j) is input to the resolution converter 15 and converted to a spatial resolution having the same size as that on the encoding side, and is output as f (x, y). Note that as a spatial resolution conversion method from f ′ (i, j) to f (x, y), an interpolation method or a simple interpolation method can be used.
[0018]
For example, when p1 = p2 = 1, the interpolation method can convert the following equations (3) to (6).
f (x, y) = f (2i, 2j) = f '(i, j)… (3)
f (x, y) = f (2i + 1,2j) = (f '(i, j) + f' (i + 1, j)) / 2… (4)
f (x, y) = f (2i, 2j + 1) = (f '(i, j) + f' (i, j + 1)) / 2… (5)
f (x, y) = f (2i + 1, 2j + 1) = (f '(i, j) + f' (i + 1, j) + f '(i, j + 1) + f' ( i + 1, j + 1)) / 4 ... (6)
Note that x, y = 0, 1, 2,..., N−1.
[0019]
In the simple interpolation method, it is possible to use transformations such as the following equations (7) to (10).
f (x, y) = f (2i, 2j) = f '(i, j)… (7)
f (x, y) = f (2i + 1,2j) = f '(i, j)… (8)
f (x, y) = f (2i, 2j + 1) = f '(i, j)… (9)
f (x, y) = f (2i + 1, 2j + 1) = f '(i, j)… (10)
On the other hand, the motion vector information b decoded by the variable length decoder 11 is input to the motion compensated predictor 17. The motion compensation predictor 17 loads the corresponding image information from the frame memory 18 according to the input motion vector information b, and outputs it from the motion compensation predictor 17 as a motion compensated prediction image c (x, y).
[0020]
The encoding mode information c decoded by the variable length decoder 11 controls the switch means 19. When the coding mode is the in-plane coding mode, the switch means 19 is turned off, and the output f (x, y) from the resolution converter 15 is not added by the adder 16 and is directly output as a decoded image output r (x , y) and is stored in the frame memory 18.
[0021]
On the other hand, when the coding mode is other than the in-plane coding mode, the switch means 19 is turned on, and the output f (x, y) from the resolution converter 15 is added by the adder 16 to the motion compensated prediction image c (x, y). ) Are added and output as decoded image output r (x, y) and stored in the frame memory 18.
[0022]
As described above, according to this embodiment, the small-basis inverse discrete cosine transformer 14 has the horizontal and vertical base sizes N / 2 scaled by the scaler 13. ^p1 , N / 2 ^p2 Is used to perform inverse discrete cosine transform, so that the processing speed can be greatly improved as compared with the conventional one.
[0023]
Next, a second embodiment of the present invention will be described with reference to FIG. In this embodiment, decoding is performed using low-pass filter processing and non-zero coefficient inverse discrete cosine transform processing.
[0024]
The encoded data is first input to the variable length decoder 11. Since the operations of the variable length decoder 11 and the inverse quantizer 12 are the same as or equivalent to those of the first embodiment, description thereof is omitted. The discrete cosine transform coefficient F (u, v) output from the inverse quantizer 12 is input to the low-pass filter 21, and the coefficient data is filtered.
[0025]
The filtering method can be realized by leaving only the low frequency coefficient in the discrete cosine transform coefficient F (u, v) having a size of N × N. If the coefficient after the low-pass filtering is F ′ (u, v), it is possible to use filtering as in the following equations (11) and (12).
When 0 ≤ u ≤ b1 and 0 ≤ v ≤ b2, F '(u, v) = F (u, v)… (11)
When u> b1 or v> b2, F '(u, v) = 0… (12)
Here, u, v = 0, 1, 2,..., N−1. B1 and b2 are integers of N or less and are filtering parameters.
[0026]
It is also possible to perform filtering using the following equations (13) and (14).
When b2 u + b1 v ≤ b1 b2, F '(u, v) = F (u, v)… (13)
When b2 u + b1 v> b1 b2, F '(u, v) = 0… (14)
The discrete cosine transform coefficient F ′ (u, v) after the low-pass filtering process is input to the non-zero coefficient inverse discrete cosine transformer 22 having the same size basis as that of the encoding side. The non-zero coefficient inverse discrete cosine transformer 22 inversely transforms the discrete cosine transform coefficient F ′ (u, v) and outputs an image f (x, y). Here, x, y = 0, 1, 2,..., N−1. Further, the basis of the non-zero coefficient inverse discrete cosine transformer 22 is the same size N × N as the discrete cosine transform on the encoding side, and u and v are coefficients b1 and b2 or less, respectively, and the equation of FIG. Inverse discrete cosine transform is performed by (15) and (16). The basis of the non-zero coefficient inverse discrete cosine transformer 22 is the same size as the discrete cosine transform on the encoding side. However, a large amount of data with F ′ (u, v) = 0 is obtained by filtering by the low-pass filter 21. As a result, the processing speed of the non-zero coefficient inverse discrete cosine transformer 22 is greatly improved.
[0027]
Instead of the inverse discrete cosine transform using the equations (15) and (16), butterfly computation may be used. FIG. 3 shows an example of a conventional fast inverse discrete cosine transform using the butterfly operation when filtering is not performed. BGLee, “A new algorithm to compute the discrete cosine transform,” IEEE Trans.Acoust. , Speech, and Signal Processing, vol. ASSP-32, pp / 1243-1245, Dec. 1984. The figure shows inverse discrete cosine processing for a one-dimensional signal when N = 8. In addition, the code | symbol in a figure has the meaning of Formula (17) of FIG. According to the process of FIG. 3, the inverse discrete cosine transform process is performed on the 8-point input of F ′ (0), F ′ (1),..., F ′ (7), and f (0), f ( 1),..., F (7) are obtained.
[0028]
However, after the filtering of this embodiment, the butterfly operation is as shown in FIG. FIG. 4 shows non-zero coefficient inverse discrete cosine transform by butterfly calculation when the filtering parameter b1 is 4 (b1 = 4). In the present embodiment, only four points F ′ (0) to F ′ (3) are input and the inverse discrete cosine transform process is performed, and f (0), f (1),..., F (7) 8 One pixel data is restored. In this way, in contrast to the conventional inverse discrete cosine transform of N point input, in this embodiment, only the inverse discrete cosine transform calculation corresponding to the non-zero low-frequency b point input signal has to be performed. Is greatly improved.
[0029]
The motion vector information decoded by the variable length decoder 11 is input to the motion compensation predictor 17 as in the first embodiment. The motion compensation predictor 17 loads the corresponding image information from the frame memory 18 according to the input motion vector information, and outputs it from the motion compensator 17 as a motion compensated prediction image c (x, y).
[0030]
The coding mode information decoded by the variable length decoder 11 also controls the switch means 19 as in the first embodiment. When the coding mode is the in-plane coding mode, the switch means 19 is turned off, and the output f (x, y) from the inverse discrete cosine transformer 22 is not added by the adder 16 and is output as the decoded image output r. It is output as (x, y) and stored in the frame memory 18.
[0031]
On the other hand, when the coding mode is other than the in-plane coding mode, the switch means 19 is turned on, and the output f (x, y) from the non-zero coefficient inverse discrete cosine transformer 22 is added to the motion compensated prediction image by the adder 16. c (x, y) is added and output as a decoded image output r (x, y) and stored in the frame memory 18.
[0032]
As described above, according to the present embodiment, the base of the non-zero coefficient inverse discrete cosine transformer 22 has the same size as that of the discrete cosine transform on the encoding side, but F ′ is filtered by the low-pass filter 21. Since a large amount of data with (u, v) = 0 exists, the substantial processing amount of the non-zero coefficient inverse discrete cosine transformer 22 is greatly reduced, and the processing speed is greatly improved. Become.
[0033]
Next, a third embodiment will be described with reference to FIG. This embodiment further improves the processing speed of the conventional method of decoding only images encoded in the intra-screen encoding mode and not decoding images encoded other than intra-screen encoding. It is what I did.
[0034]
The encoded data is input to the variable length decoder 11. The variable length decoder 11 decodes the encoding mode, and the quantized discrete cosine transform coefficient a, the encoding mode information c, and the like are decoded. When the encoding mode is the in-plane encoding mode, the switch unit 11 a is closed, and the quantized discrete cosine transform coefficient a is output from the variable length decoder 11 to the inverse quantizer 12. The quantized discrete cosine transform coefficient a input to the inverse quantizer 12 is inversely quantized to output a discrete cosine transform coefficient F (u, v). The discrete cosine transform coefficient F (u, v) is input to the scaler 13, and the coefficient data is scaled.
[0035]
As for the scaling method, each discrete cosine transform coefficient is changed in accordance with the equation given by the equation (1). In addition, scaling as in the above equation (2) can also be used.
[0036]
The discrete cosine transform coefficient F ′ (u, v) after the scaling processing is input to the small basis inverse discrete cosine transformer 14. In the small-basis inverse discrete cosine transformer 14, the horizontal and vertical base sizes are N / 2. ^p1 , N / 2 ^p2 The discrete cosine transform coefficient F ′ (u, v) is inversely transformed by the inverse discrete cosine transform, and an image f ′ (i, j) is output. The image f ′ (i, j) is input to the resolution converter 15 and converted to a spatial resolution having the same size as that on the encoding side, and is output as f (x, y). As a spatial resolution conversion method from f ′ (i, j) to f (x, y), the same interpolation method or simple interpolation method as in the first embodiment can be used.
The output f (x, y) from the resolution converter 15 is output as the decoded image output r (x, y) as it is.
[0037]
According to this embodiment, the processing speed of an image encoded in the in-screen encoding mode can be greatly improved.
[0038]
Next, a fourth embodiment of the present invention will be described with reference to FIG. Similarly to the third embodiment, this embodiment also improves the processing speed of an image encoded in the intra-screen encoding mode.
[0039]
The encoded data is input to the variable length decoder 11. The variable length decoder 11 decodes the encoding mode, and the quantized discrete cosine transform coefficient a, the encoding mode information c, and the like are decoded. When the encoding mode is the in-plane encoding mode, the switch unit 11 a is closed, and the quantized discrete cosine transform coefficient a is output from the variable length decoder 11 to the inverse quantizer 12. The quantized discrete cosine transform coefficient a input to the inverse quantizer 12 is inversely quantized to output a discrete cosine transform coefficient F (u, v). The discrete cosine transform coefficient F (u, v) is input to the low-pass filter 21, and the coefficient data is filtered.
[0040]
As a filtering method, if the coefficient after low-pass filtering is F ′ (u, v), it is possible to use the filtering as in the equations (11) and (12). It is also possible to perform filtering using the equations (13) and (14). This filtering makes it possible to leave only the low-frequency coefficients in the discrete cosine transform coefficient F (u, v) of size N × N.
[0041]
The discrete cosine transform coefficient F ′ (u, v) after the low-pass filtering process is input to the non-zero coefficient inverse discrete cosine transformer 22. The basis of the non-zero coefficient inverse discrete cosine transformer 22 is the same size N × N as that of the discrete cosine transform on the encoding side, and u and v are coefficients b1 and b2 or less, respectively, and the equation (15), Perform inverse discrete cosine transform by (16). Further, as in the second embodiment, the butterfly calculation of FIG. 4 may be used.
[0042]
The non-zero coefficient inverse discrete cosine transformer 22 inversely transforms the discrete cosine transform coefficient F ′ (u, v) and outputs an image f (x, y). Here, x, y = 0, 1, 2,..., N−1. The output f (x, y) from the inverse discrete cosine transformer 22 is output as the decoded image output r (x, y) as it is.
[0043]
According to this embodiment, as in the third embodiment, the processing speed of an image encoded in the intra-screen encoding mode can be greatly improved.
[0044]
Next, a fifth embodiment of the present invention will be described with reference to FIG. In the figure, 11a indicates a switch means, and other reference numerals indicate the same as or equivalent to those in FIG. In this embodiment, the speed of decoding processing of a screen subjected to in-plane encoding and one-way predictive encoding is improved.
[0045]
The encoded data is input to the variable length decoder 11, and the variable length decoder 11 decodes the quantized discrete cosine transform coefficient a, motion vector information b, encoding mode information c, and the like. The encoding mode information c controls the operation of the switch means 11a, 19.
[0046]
The encoding mode information c is such that the switch means 11a is turned on when the encoding mode is the in-plane encoding mode and the unidirectional predictive coding screen mode, and the switch 19 is turned off when the encoding mode is in the in-plane encoding mode. Switch 19 is turned on in the screen mode. Also, it is turned off in other than these encoding modes. As a result, when the encoding mode is the in-plane encoding mode or the unidirectional predictive encoding screen mode, the quantized discrete cosine transform coefficient a is input to the inverse quantizer 12 and is the same as described in the first embodiment. As a result, the decoded image output r (x, y) is finally obtained. According to this embodiment, only the in-plane encoding screen and the unidirectional predictive encoding screen are decoded, and the other encoding screens are not decoded. In addition to improving the decoding speed by this thinning, the first embodiment Since the processing speed by the small basis inverse discrete cosine transformer 14 described in (1) can be improved, the processing speed can be improved more than the processing speed of the first embodiment.
[0047]
Next, a sixth embodiment of the present invention will be described with reference to FIG. In the figure, reference numeral 11a denotes a switch means, and other reference numerals are the same as or equivalent to those in FIG. In this embodiment, the speed of decoding processing of a screen subjected to in-plane encoding and one-way predictive encoding is improved.
[0048]
The encoded data is input to the variable length decoder 11, and the variable length decoder 11 decodes the quantized discrete cosine transform coefficient a, motion vector information b, encoding mode information c, and the like. The encoding mode information c controls the operation of the switch means 11a, 19.
[0049]
The encoding mode information c is such that the switch means 11a is turned on when the encoding mode is the in-plane encoding mode and the unidirectional predictive coding screen mode, and the switch 19 is turned off when the encoding mode is in the in-plane encoding mode. Switch 19 is turned on in the screen mode. Also, it is turned off in other than these encoding modes. As a result, when the encoding mode is the in-plane encoding mode or the unidirectional predictive encoding screen mode, the quantized discrete cosine transform coefficient a is input to the inverse quantizer 12 and is the same as described in the second embodiment. As a result, the decoded image output r (x, y) is finally obtained. According to this embodiment, similarly to the fifth embodiment, it is possible to improve the decoding speed of the in-plane encoding screen and the unidirectional prediction encoding screen.
[0050]
Next, a seventh embodiment of the present invention will be described with reference to FIG. In the figure, 50 is an inverse discrete cosine transformer, 51 and 52 are switching units, and the other symbols are the same as or equivalent to those in FIG. In this embodiment, since an intra-encoded image is the basis for decoding an image encoded elsewhere, the intra-encoded image is decoded in its entirety and encoded otherwise. The speed of decoding the converted image is improved.
[0051]
The encoded data is input to the variable length decoder 11, and the variable length decoder 11 decodes the quantized discrete cosine transform coefficient a, motion vector information b, encoding mode information c, and the like.
[0052]
The quantized discrete cosine transform coefficient a is input to the inverse quantizer 12 and the motion vector information is input to the motion compensated predictor 17. The quantized discrete cosine transform coefficient a input to the inverse quantizer 12 is inversely quantized to output a discrete cosine transform coefficient F (u, v).
[0053]
When the coding mode is the in-plane coding mode, the switch 51 is connected to the terminal s1, the switch 52 is connected to the terminal s3, the switch 19 is turned off, and the quantized discrete cosine transform coefficient a is reversed. Input to the discrete cosine transformer 50. The inverse discrete cosine transformer 50 performs inverse discrete cosine transformation with the same base size N × N as that on the encoding side, outputs f (x, y), and outputs it as a decoded image output r (x, y) as it is. Accumulated in the frame memory 18.
[0054]
On the other hand, when the coding mode is other than the in-plane coding mode, the switch 51 is connected to the terminal s2, the switch 52 is connected to the terminal s4, the switch 19 is turned on, and the discrete cosine transform coefficient is The coefficient data is input to the scaler 13 and scaled. The discrete cosine transform coefficient F ′ (u, v) after the scaling processing is input to the small basis inverse discrete cosine transformer 14, and then the output is input to the resolution converter 15. The operations of the scaler 13, the small basis inverse discrete cosine converter 14, and the resolution converter 15 and the operations of the other components are the same as those in the first embodiment, and thus the description thereof is omitted.
[0055]
According to this embodiment, it is possible to simultaneously improve the image quality of the decoded image and the decoding processing speed.
[0056]
Next, an eighth embodiment of the present invention will be described with reference to FIG. In the figure, 60 is an inverse discrete cosine transformer, 61 and 62 are switching units, and the other symbols are the same as or equivalent to those in FIG. In this embodiment, as in the seventh embodiment, the image encoded in the plane is decoded in a complete form, and the decoding speed is improved for images encoded in other cases. It is a thing.
[0057]
In this embodiment, when the coding mode is the in-plane coding mode, the switch 61 is connected to the terminal s1, the switch 62 is connected to the terminal s3, the switch 19 is turned off, and the quantized discrete cosine. The transform coefficient a is input to the inverse discrete cosine transformer 60. The inverse discrete cosine transformer 60 performs inverse discrete cosine transformation with the same base size N × N as that on the encoding side, outputs f (x, y), and outputs it as a decoded image output r (x, y) as it is. Accumulated in the frame memory 18.
[0058]
On the other hand, when the coding mode is other than the in-plane coding mode, the switch 61 is connected to the terminal s2, the switch 62 is connected to the terminal s4, the switch 19 is turned on, and the discrete cosine transform coefficient F (u, v) is input to the low-pass filter 21 to filter the coefficient data. The filtered data is then input to a non-zero coefficient inverse discrete cosine transformer 22. Since the operations of the low-pass filter 21 and the non-zero coefficient inverse discrete cosine transformer 22 and the operations of the other components are the same as those of the second embodiment, description thereof will be omitted.
[0059]
According to this embodiment, as in the seventh embodiment, it is possible to simultaneously improve the image quality of the decoded image and the decoding processing speed.
[0060]
Next, the function of the decryption device of each embodiment described above can be realized by software (program), and the software can be recorded on a portable recording medium such as an optical disk, a floppy disk, or a hard disk.
[0061]
FIG. 13 shows an example of a program recorded on the recording medium. FIG. 13 (a) shows the contents of the recording medium for executing the functions of the embodiment of FIG. 1, and FIG. b) shows the contents of the recording medium for executing the functions of the embodiment of FIG.
[0062]
13A includes a variable length decoding function 111, an inverse quantization function 121, a scaling function 131, a small basis inverse discrete cosine transform function 141, a resolution conversion function 151, an addition function 161, and a motion compensated prediction. A decryption program of function 171 is recorded.
[0063]
The recording medium 200 in FIG. 13B includes a variable length decoding function 111, an inverse quantization function 121, a low-pass filter function 211, a non-zero coefficient inverse discrete cosine transform function 221, an addition function 161, and a motion compensation prediction function 171. A decryption program is recorded. It should be noted that the functions recorded on the recording medium may not be all of the above functions but only the main functions.
[0064]
FIG. 14 is a block diagram showing a hardware configuration of a computer that reads a decoding program recorded on the portable recording medium and executes a decoding function. The computer 100 includes a recording medium 200 on which the decoding program is recorded, a reading device 101 that reads the decoding program from the recording medium 200, a CPU 102 that executes the decoding program, and a ROM 103 that stores various data. A RAM 104 that stores calculation parameters, an input device 102 such as a keyboard and a mouse, an output device 105 such as a display and a printer, and a bus 106 that connects each unit of the device.
[0065]
The CPU 102 reads the decoding processing program recorded on the recording medium 200 via the reading device 101, and then executes the decoding processing by executing the decoding processing program. As the frame memory 18 used for executing the decryption processing program, a partial area of the RAM 104 or a partial area of the hard disk (not shown) can be used. The recording medium 200 includes a transmission medium that temporarily records and holds data, such as a network.
[0066]
The encoded data to be decoded may be stored in advance in a memory such as the hard disk or may be taken into the computer 100 from a network not shown.
[0067]
【The invention's effect】
As described above, according to the present invention, the decoding process is performed using the inverse discrete cosine transform of the small base size, the low-pass filter of the discrete cosine transform coefficient and the non-zero coefficient inverse discrete cosine transform. Also, it becomes possible to perform moving image decoding at high speed. In addition, it is possible to provide a recording medium storing a program capable of performing moving image decoding processing at high speed.
[0068]
For example, the moving picture encoding method H.264 in ITU. In the simulation experiment by H.263, when the apparatus of FIG. 1 (first embodiment) is used, decoding can be performed 30% or more faster than the conventional case in which inverse discrete cosine transform having the same base size as that of the encoding side is performed. It was.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a first embodiment of the present invention.
FIG. 2 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.
FIG. 3 is an explanatory diagram showing an example of fast inverse discrete cosine transform using butterfly computation.
FIG. 4 is an explanatory diagram of butterfly computation applied to the second embodiment of the present invention.
FIG. 5 is a block diagram showing a configuration of a third exemplary embodiment of the present invention.
FIG. 6 is a block diagram showing a configuration of a fourth embodiment of the present invention.
FIG. 7 is a block diagram showing a configuration of a fifth exemplary embodiment of the present invention.
FIG. 8 is a block diagram showing a configuration of a sixth embodiment of the present invention.
FIG. 9 is a block diagram showing a configuration of a seventh embodiment of the present invention.
FIG. 10 is a block diagram showing a configuration of an eighth embodiment of the present invention.
FIG. 11 is a diagram illustrating mathematical expressions.
FIG. 12 is a block diagram showing a configuration of a conventional apparatus.
FIG. 13 is a diagram showing an outline of a program recorded on a recording medium.
FIG. 14 is a block diagram showing a configuration of a computer that executes a decryption program recorded on a recording medium of the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 11 ... Variable length decoder, 12 ... Inverse quantizer, 13 ... Scaling device, 14 ... Small base inverse discrete cosine transformer, 15 ... Resolution converter, 16 ... Adder, 17 ... Motion compensation predictor, 18 ... Frame Memory, 19 ... switch means, 21 ... low pass filter, 22 ... non-zero coefficient inverse discrete cosine transformer, 50, 60 ... inverse discrete cosine transformer.

Claims (8)

In a decoding device of encoded moving image data for decoding moving image data compressed and encoded using motion compensated prediction and discrete cosine transform,
Inverse quantized discrete cosine transform coefficients F (u, v) and, using equation (1) or (2) below, the base of the smaller than the discrete cosine transform basis on the encoding side, and the DC component only Means for changing to a base conversion coefficient F ′ (u, v) not including
Conversion coefficient F '(u, v) which has been changed by said means, inversely transformed using a horizontal inverse discrete cosine transform base size in the vertical direction using a base of ^{N / 2 p1, N / 2} p2 Means,
An apparatus for decoding encoded moving image data, comprising: means for converting the inverse discrete cosine transformed image data into the same size as that on the encoding side.

F ′ (u, v) = F (u, v) / {(2 ^{p1 / 2} ) × (2 ^{p2 / 2} )} (2)
Here, u and v are the coordinates of the discrete cosine transform coefficients in the horizontal direction and the vertical direction, p1 and p2 are parameters for determining the base sizes in the horizontal and vertical directions, respectively, and N × N is the discrete cosine transform on the encoding side. Is the base size of. In the decoding apparatus of the encoded moving image data according to claim 1,
An apparatus for decoding encoded moving image data, wherein only an image subjected to in-plane encoding is decoded. In the decoding apparatus of the encoded moving image data according to claim 1,
Means for performing motion compensation prediction with block data of the same size as the encoding side, and restoring the image block data subjected to inverse discrete cosine transform to the same block size as the encoding side into moving image data;
A decoding apparatus for encoded moving image data, further comprising means for storing the restored moving image data for the motion compensation prediction. In the decoding apparatus of the encoded moving image data according to claim 3,
A decoding apparatus for encoded moving image data, wherein only an image subjected to in-plane encoding and an image subjected to one-way predictive encoding are decoded. In the decoding apparatus of the encoded moving image data according to claim 3,
Inverse discrete cosine transform with the same block size as the encoding side, connected via switching means and inverse transforming means using inverse discrete cosine transformation using a basis smaller than the discrete cosine transform basis on the encoding side Further comprising means for
In-plane encoded image is decoded by means of inverse discrete cosine transform with the same block size as the encoding side,
The encoded image other than the in-plane encoded image is subjected to inverse transform processing using inverse discrete cosine transform using a basis smaller than the discrete cosine transform basis on the encoding side, and decoded by performing the motion compensation prediction. An apparatus for decoding encoded moving image data, characterized in that: In the decoding apparatus of the encoding moving image data in any one of Claim 1 thru | or 5,
The means for changing the inversely quantized discrete cosine transform coefficient to a transform coefficient for a base smaller than the discrete cosine transform base on the encoding side is a scaling processing means. Decoding device. Inverse quantized discrete cosine transform coefficients F (u, v) and, by (1) or (2) below, smaller than the discrete cosine transform basis on the encoding side, and includes a base only the DC component Changing to a non-basic transform coefficient F ′ (u, v) ;
The transform coefficient F which is changed by the step '(u, v), is inversely transformed using a horizontal inverse discrete cosine transform base size in the vertical direction using a base of ^{N / 2 p1, N / 2} p2 Process,
Converting the inverse discrete cosine transformed image data to the same size as the encoding side,
A computer-readable recording medium on which a program for causing a computer to execute is recorded.

F ′ (u, v) = F (u, v) / {(2 ^{p1 / 2} ) × (2 ^{p2 / 2} )} (2)
Here, u and v are the coordinates of the discrete cosine transform coefficients in the horizontal direction and the vertical direction, p1 and p2 are parameters for determining the base sizes in the horizontal and vertical directions, respectively, and N × N is the discrete cosine transform on the encoding side. Is the base size of. The recording medium according to claim 7,
Furthermore, a computer-readable recording medium on which a program for a process of restoring a moving image by performing motion compensation prediction using block data having the same size as that of the encoding side is recorded.

Images