JP2001197492A - Encoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce encoding generation bit amount, without deteriorating image quality in real time. SOLUTION: An encoding device is provided with a first movement vector detector 101 comprising an image data storage unit, an encoder 102 comprising a storage unit and a second movement vector detector, a frame buffer 103 for delaying image data to be inputted to the first movement vector detector 101 for the portion of one image and a maximum differential value extractor 104 for extracting the maximum differential value among movement vector differential values by each macro block in one picture detected by the detector 101. Then the maximum differential value among the movement vectors by macro block in one image is extracted, a minimum fcode which can express the maximum differential value is provided to the encoder 102 and, then encoding is performed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an encoding apparatus for encoding image data using a moving image compression algorithm such as MPEG.

[0002]

2. Description of the Related Art As shown in FIG. 10, a moving image compression encoding method such as MPEG employs a frame memory 1001, a motion detection unit 1002, a motion compensation prediction unit 1003 for input image data, an input image data Unit 1004 for calculating a difference between the image data and a predicted value for the image data, DC
T transform processing section 1005, quantization processing section 1006, variable length coding section 1007, inverse quantization section 1008, inverse DCT processing section 1009, input image data is subjected to DCT, quantization,
It is composed of a frame memory 1010 for image data after inverse quantization and inverse DCT processing, and predictively encodes temporally past or future pictures in order to reduce temporal redundancy.

The input image data is obtained by detecting a motion vector by a motion detecting unit 1002 by referring to a frame memory 1001 in which image data of past and future pictures is stored. , And the difference from the input image is calculated by the difference unit 1004 to reduce temporal redundancy. After that, DC
Through the T processing unit 1005, the quantization unit 1006, and the variable-length coding unit 1007, compression coding is performed with spatial redundancy reduced.

[0004]

However, even if the above-described encoding is performed, the amount of generated bits may be large.
Further reduction in the amount of generated bits is required. Here, encoding parameters that affect the amount of generated bits include:
Although there is an fcode, this fcode is set uniformly at the time of encoding although it is allowed to be set for each picture, and often includes redundant bits.

[0005] The present invention has been made to solve such a problem, and has been made in consideration of a real-time encoding parameter.
An object of the present invention is to provide an encoding device that can reduce the amount of generated bits without deteriorating image quality by optimizing and encoding fcode.

[0006]

According to a first aspect of the present invention, an encoding apparatus includes an encoder including a first motion vector detector, a second motion vector detector, and the first motion vector detector. And a frame buffer capable of delaying the image data input to one picture by one picture, and a maximum value among the difference values of the motion vectors for each macroblock in one picture detected by the first motion vector detector. Has a maximum difference value extractor for extracting the difference value of.

[0007] Also, an encoder according to a second aspect of the present invention includes n encoders including a first motion vector detector and a second motion vector detector, each of which differs only in fcode setting.
A maximum difference value extractor for extracting a maximum difference value among difference values of motion vectors for each macroblock in one picture detected by the first motion vector detector;
Compare the n generated bit amount counters that count the generated bit amounts of the respective encoded data encoded by the encoders with the generated bit amounts counted by the n generated bit amount counters, and And a coded data selector for selecting and outputting coded data of the coder selected by the generated bit amount comparator. I have.

[0008] The encoding apparatus according to a third aspect of the present invention comprises an encoder including a motion vector detector, and a difference value of a motion vector for each macroblock within one picture detected by the motion vector detector. Among them, it has a maximum difference value extractor for extracting a maximum difference value.

Further, a coding apparatus according to a fourth aspect of the present invention includes an encoder including a motion vector detector and an encoded data storage, and an encoder for each macroblock in one picture detected by the motion vector detector. A maximum difference value extractor that extracts the maximum difference value among the difference values of the motion vectors, and an fcode candidate storage that stores fcode candidates that can represent the maximum difference value extracted by the maximum difference value extractor And a generated bit amount counter for counting the number of encoded bits of one picture, and a generated bit amount comparator for comparing the generated bit amount for each picture.

The encoding method of the present invention extracts a maximum difference value among motion vector difference values for each macroblock in one picture, and extracts a minimum difference value capable of expressing the maximum difference value. It is characterized in that image data is encoded using fcode.

[0011]

Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1) In moving image compression encoding such as MPEG, a motion vector is detected in order to reduce temporal redundancy. Here, detection of a motion vector is mainly performed by block matching.

At this time, the sum of the absolute values of the pixel differences is evaluated as an evaluation value indicating the degree of correlation between the block image of interest, from which the motion vector is to be detected, and all candidate block images within the search range in the reference frame. Is calculated, and the displacement up to the candidate block having the smallest evaluation value is obtained as a motion vector.

Further, when coding a motion vector,
In a nearby macroblock in the image (a unit for obtaining a motion vector in the image), a difference value from a motion vector of a previously encoded macroblock is encoded by using a characteristic that motion vectors are similar. ing.

This difference value (delta) is defined by ISO / IEC 13818-2.
As described in 7.6.3.1, the basic vector components (mo
is calculated by scaling the motion_code by the scale factor (f) and adding a residual vector component (motion_residual) (Equation 1).

(Equation 1) delta = Sign (motion # code) * [((Abs (motion # code) -1) * f) + motion # residual + 1] Here, f is set for each picture. Is calculated from the permissible encoding parameter fcode.
Is raised to the power of (fcode-1). Usually, a uniform value is set at the start of encoding.

Also, motion # code is based on ISO / IEC 13818-2.
It is as described in Table B.10 (Figure 2). As shown in FIG. 2, motion_code is given as a variable-length code, and motion_residual is (fcode-1)
With the number of bits.

Thus, delta is f, motion # code, m
With otion_residual, a wide range of motion vectors can be encoded. Furthermore, Table of ISO / IEC 13818-2
7-8 shows the relationship between the fcode and the allowable motion vector range (FIG. 3). As shown in FIG. 3, the range of the motion vector that can be represented by the value of fcode can be changed. .

FIG. 1 shows an encoding apparatus according to the first embodiment of the present invention. As shown in FIG.
The encoding apparatus according to the embodiment includes a first motion vector detector 101 including an image data storage, and a second motion vector detector equivalent to the image data storage and the first motion vector detector 101. And a frame buffer 10 that can delay image data input to the first motion vector detector 101 by one picture.
3 and a maximum difference value extractor 104 for extracting the maximum delta among the difference values (delta) of the motion vectors for each macroblock in one picture detected by the first motion vector detector 101. It was done.

In this embodiment, first, the original image data is sent to the image data storage in the first motion vector detector 101, and the first motion vector detector 101 performs the motion vector detection. It is assumed that data of several pictures necessary for encoding is stored in the image data storage.

In the first motion vector detector 101, a motion vector is sequentially extracted for each macroblock of the image to be coded, and a difference value from the motion vector extracted in the immediately preceding macroblock is calculated for each macroblock. It is calculated for each.

The difference value calculated for each macroblock is transferred to the maximum difference value extractor 104, and the detection of motion vectors in all macroblocks of the image to be encoded is completed, and the calculation of the difference value is completed. At this stage, the maximum difference value extractor extracts the maximum difference value from the encoding target image, and utilizes the relationship shown in FIG. 3 to express the minimum difference value that can express the maximum difference value. Deriving fcode.

The derived minimum fcode is calculated by the encoder 1
02, the encoder 102 encodes the same encoding target image delayed by the frame buffer 103 using the minimum fcode. The encoded data encoded by the encoder 102 is output as encoded data of the encoding device.

For example, let us consider a case where the fcode which has been uniformly set in a normal encoding apparatus is "6" and the maximum difference value of a certain encoding target image is "+30". In this case, when the encoding device according to the present embodiment is used, the maximum difference value extractor 1
04 is derived from the relationship shown in FIG. 3 as the minimum fcode that can express “+30”, that is, “3” as the minimum fcode that can be set. So for this picture, fcode
Is set to "3", so that the number of bits (fcode-1) allocated to motion_residual in all macroblocks of the picture is changed from "5" in a normal coding apparatus to "2". Can be reduced.

With the above configuration, generally, moti
By reducing the number of bits assigned to on_residual, the possibility of reducing the number of generated bits without deteriorating image quality increases.

(Embodiment 2) FIG. 4 shows an encoding apparatus according to a second embodiment of the present invention. As shown in FIG. 4, an encoding apparatus according to a second embodiment of the present invention includes a first motion vector detector 401 including an image data storage, an image data storage and a first motion vector detector N encoders 402 each including a second motion vector detector equivalent to 401 and differing only in fcode setting, and for each macroblock in one picture detected by the first motion vector detector 401 Of the motion vector difference values
A maximum difference value extractor 403 for extracting a maximum difference value;
The n generated bit amount counters 404 for counting the generated bit amounts of the respective pieces of encoded data encoded by the encoders 402 are compared with the counted generated bit amounts to find the minimum generated bit amount. A generated bit amount comparator 405 for selecting an encoder, and an encoded data selector 40 for selecting and outputting encoded data of the selected encoder
6.

In this embodiment, the first motion vector detector 401 sequentially detects a motion vector for each macroblock of an image to be coded, and calculates a difference from the motion vector detected in the immediately preceding macroblock. The value is calculated for each macroblock.

The difference value calculated for each macroblock is transferred to the maximum difference value extractor 403, and the detection of motion vectors in all macroblocks of the image to be encoded is completed, and the calculation of the difference value is completed. At this stage, the maximum difference value extractor 403 extracts the maximum difference value from the image to be encoded, and uses the relationship shown in FIG. 3 to express the fcode that can express the maximum difference value. Is derived. For example, when the maximum difference value is “+30”, the maximum difference value extractor 403 uses “3” to “3” from the relationship shown in FIG.
9 "is derived as an fcode candidate.

At the same time, the same number of encoders 402, which differ only in the fcode setting, perform encoding on the same encoding target image, and the amount of bits generated by encoding corresponds to the corresponding amount of generated bits. It is calculated by the counter 404.

Thereafter, the generated bit amount comparator 405 selects an encoder having the smallest generated bit amount among the encoders that have performed the encoding based on the derived fcode candidates, and selects the encoded data. Notify the container 406. Finally, the encoded data selector 406 outputs the encoded data of the encoder selected by the generated bit amount comparator 405 as the encoded data of the encoding device.

Here, since (Equation 1) is used to express a motion vector, in the first embodiment,
In rare cases, even if the number of bits allocated to motion_residual can be reduced by optimizing fcode, the number of bits allocated to motion_code increases, and the amount of bits generated per picture may instead increase.

For example, let us consider a case where the fcode, which has been uniformly set in a normal encoding apparatus, is "9" and the maximum difference value of an image to be encoded is "+521". In this case, when the encoding apparatus according to the first embodiment is used, the maximum difference value extractor 104 can set the minimum fcode that can express “+521”, that is, “8” from the relationship shown in FIG. Derived as the minimum fcode.

Therefore, by setting the fcode to "8", the number of bits (fcode-1) allocated to motion_residual in this macroblock is changed from "8" to "7" in the ordinary coding apparatus. Can be reduced to "mo
The number of bits allocated to the action_code is increased from "5" to "8", and the generated bit amount of this macro block is increased by 2 bits (d
elta = [((3-1) * 256) + 8 + 1] to delta = [((5-1) * 1
As a result, the motion_code is changed from “00010” representing “3” to “0000101” representing “5” as shown in FIG.
This is changed to "0", which increases by 3 bits).

However, according to the present embodiment, since the number of generated bits is actually encoded for each fcode candidate, the generated bit amount can be reliably reduced.

With the above configuration, by reducing redundant bits, the amount of generated bits can be reduced without deteriorating image quality.

(Embodiment 3) FIG. 5 shows an encoding apparatus according to a third embodiment of the present invention. As shown in FIG. 5, an encoding device according to a third embodiment of the present invention includes an encoder 501 including an image data storage and a motion vector detector,
Among the difference values of the motion vectors for each macroblock in one picture detected by the motion vector detector,
It has a maximum difference value extractor 502 for extracting the maximum difference value.

In the present embodiment, first, original image data is sent to an image data storage in the encoder 501, and a motion vector is detected in a motion vector detector in the encoder 501. It is assumed that data of several pictures necessary for encoding is stored in the image data storage.

The motion vector detector in the encoder 501 sequentially detects a motion vector for each macroblock of the image to be coded, and calculates a difference value from the motion vector detected in the immediately preceding macroblock. It is calculated for each macroblock.

The difference value calculated for each macroblock is transferred to the maximum difference value extractor 502, and the detection of motion vectors in all macroblocks of the image to be encoded is completed, and the calculation of the difference value is completed. At this stage, the maximum difference value extractor 502 extracts the maximum difference value from the image to be encoded and uses the relationship shown in FIG. 3 to extract the minimum difference value that can express the maximum difference value. Deriving fcode of

The derived minimum fcode is calculated by the encoder 5
01, and the encoder 501 encodes the same encoding target image using the minimum fcode. The coded data obtained by this coding is output as coded data of the coding apparatus according to the present embodiment.

Here, when encoding is performed on an image smaller than the image size that can be processed by the encoder 501, as shown in FIG. 6, there is a margin in the allowable processing time for realizing real-time encoding. Therefore, the above-described encoding can be realized using the extra time.
Similarly, when the processing capability of the encoder is sufficiently high,
If there is enough processing time, the above-described encoding can be realized.

In general, when there is a margin for the processing time for realizing the real-time encoding, the motion_residu
By reducing the number of bits allocated to al, the possibility of reducing the number of generated bits without deteriorating image quality increases.

(Embodiment 4) FIG. 7 shows an encoding apparatus according to a fourth embodiment of the present invention. As shown in FIG. 7, an encoding device according to a fourth embodiment of the present invention includes an encoder 701 including an image data storage, a motion vector detector, and an encoded data storage, and a motion vector detector. The maximum difference value extractor 702 that extracts the maximum difference value among the motion vector difference values for each macroblock in one detected picture, and the maximum difference value extracted by the maximum difference value extractor 702 are An fcode candidate storage 703 that stores fcode candidates that can be expressed, a generated bit amount counter 704 that counts the number of encoded bits generated for one picture, and a generated bit amount comparator that compares the generated bit amount for each picture 705.

In this embodiment, first, the original image data is sent to the image data storage in the encoder 701, and the motion vector is detected in the motion vector detector in the encoder 701. In the image data storage,
It is assumed that data of several pictures required for encoding is stored.

The motion vector detector in the encoder 701 sequentially detects a motion vector for each macroblock of the image to be coded, and calculates a difference value from the motion vector detected in the immediately preceding macroblock. It is calculated for each macroblock.

The difference value calculated for each macroblock is transferred to the maximum difference value extractor 702, and the detection of motion vectors in all macroblocks of the image to be encoded is completed, and the calculation of the difference value is completed. At this stage, the maximum difference value extractor 702 extracts the maximum difference value from the encoding target image, and can express the maximum difference value using the relationship shown in FIG. Is derived and stored in the fcode candidate storage 703.

For example, when the maximum difference value is “+30”, the maximum difference value extractor 702 derives “3” to “9” as fcode candidates from the relationship shown in FIG. It is stored in the candidate storage unit 703.

Thereafter, one of the stored fcode candidates is f
Provided from the code candidate storage 703 to the encoder 701,
Encoding is performed on the same encoding target image.

When this encoding is completed, the encoder 701
Sends a one-picture encoding end signal to the generated bit amount counter 704, the generated bit amount counter 704 counts the generated bit amount per picture, transfers it to the generated bit amount comparator 705 and holds the value. , The encoded data is stored in the determined area of the encoded data storage shown in FIG.

Similarly, when encoding of one picture is completed,
When the one-picture encoding end signal is sent from the encoder 701 to the fcode candidate storage 703, the next fcode candidate is fco
After being provided from the de candidate storage unit 703 to the encoder 701 and encoding is performed again, the generated bit amount per picture is counted by the generated bit amount counter 704 and transferred to the generated bit amount comparator 705. In addition, in the re-encoding after the second encoding, the encoded data is stored in the temporary area of the encoded data storage shown in FIG.

At this time, if the current encoding is smaller than the amount of bits generated at the previous encoding,
The generated bit amount comparator 705 sends an encoded data update request to the encoded data storage, and the encoded data storage transfers the encoded data from the temporary area to the determined area and overwrites the data. The generated bit amount comparator 705 holds the generated bit amount in the current encoding as the minimum generated bit amount.

Finally, at the time when encoding with all fcode candidates is completed, the encoded data with the least amount of generated bits is stored in the fixed area of the encoded data storage. The encoding device according to the present embodiment outputs encoded data stored in the determined area.

Here, when encoding is performed on an image of 1 / n of the image size that can be processed by the encoder 701, as shown in FIG. 8, an allowable processing time for realizing the real-time encoding is obtained. Therefore, the above-described encoding can be realized using the extra time.

Similarly, when the processing capability of the encoder is sufficiently high, the above-described encoding can be realized if there is a margin for the processing time.

With the above configuration, by reducing redundant bits, the amount of generated bits can be reliably reduced without deteriorating image quality.

[0056]

As described above, according to the first aspect, motion_resid used to encode a difference value of a motion vector by optimizing fcode before encoding.
The number of bits allocated to the ual can be reduced, and the amount of generated bits can be reduced without deteriorating the image quality.

According to the second aspect of the present invention, the number of generated bits is actually encoded and counted for each fcode candidate that does not degrade the image quality, so that the generated bit amount can be reliably reduced. .

Further, according to the third invention, if there is a margin in the processing time for realizing the real-time encoding,
Using that extra time, optimize fcode, motion # re
By reducing the number of bits allocated to sidual, it is possible to provide a small-scale encoding device capable of reducing the amount of generated bits without deteriorating image quality.

According to the fourth aspect of the present invention, if there is a margin in the processing time for realizing the real-time encoding, the image quality is not degraded by utilizing the margin time.
Since the amount of generated bits is actually encoded for each ode candidate, the coding device can be provided with a small scale capable of reliably reducing the generated bit amount.

[Brief description of the drawings]

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

FIG. 2 is a diagram showing a variable-length code of motion_code.

FIG. 3 is a diagram showing a relationship between fcode and a range of expressible motion vectors.

FIG. 4 is a block diagram showing an encoding device according to a second embodiment of the present invention.

FIG. 5 is a block diagram showing an encoding device according to a third embodiment of the present invention.

FIG. 6 is a diagram showing a processing time in encoding according to a third embodiment of the present invention.

FIG. 7 is a block diagram illustrating an encoding device according to a fourth embodiment of the present invention.

FIG. 8 is a diagram showing processing time in encoding according to a fourth embodiment of the present invention.

FIG. 9 is a diagram showing an encoded data storage according to a fourth embodiment of the present invention.

FIG. 10 is a diagram showing a flow of moving image compression encoding.

[Explanation of symbols]

 Reference Signs List 101 motion vector detector 102 encoder 103 frame buffer 104 maximum difference value extractor 401 motion vector detector 402 n encoders 403 maximum difference value extractor 404 n generated bit amount counter 405 generated bit amount comparator 406 Encoded data selector 501 Encoder 502 Maximum difference value extractor 701 Encoder 702 Maximum difference value extractor 703 fcode candidate storage 704 Generated bit amount counter 705 Generated bit amount comparator 901 Encoded data storage

Claims (5)

[Claims] An encoder including a first motion vector detector, a second motion vector detector, and an encoder for delaying image data input to the first motion vector detector by one picture. And a maximum difference value extractor for extracting a maximum difference value among difference values of motion vectors for each macroblock in one picture detected by the first motion vector detector. Monitoring a difference value of a motion vector for each macroblock in one picture detected by the first motion vector detector by the maximum difference value extractor, and extracting a maximum difference value in the picture; An encoding device in which the encoder performs encoding using a minimum fcode capable of expressing the maximum difference value. 2. An encoder comprising a first motion vector detector and a second motion vector detector, wherein n encoders each differ only in fcode setting, and are detected by the first motion vector detector. A maximum difference value extractor for extracting a maximum difference value among motion vector difference values for each macroblock in one picture, and a maximum difference value extractor for each of the encoded data encoded by the n encoders. N generated bit amount counters for counting generated bit amounts, and generated bit amounts for comparing the generated bit amounts counted by the n generated bit amount counters and selecting an encoder having the smallest generated bit amount And a coded data selector for selecting and outputting coded data of the coder selected by the generated bit amount comparator, wherein one bit detected by the first motion vector detector is provided. The maximum difference value extractor monitors the difference value of the motion vector for each macroblock in the picture, extracts the largest difference value in the picture, and is an fcode candidate capable of expressing the largest difference value. Is provided to the generated bit amount comparator, the n encoders respectively perform coding, and the generated bit amount counter generated by the encoding is counted by the n generated bit amount counters,
The generated bit amount comparator compares the generated bit amount in the encoder coded by the fcode candidate, selects the encoder having the smallest generated bit amount, and the coded data selector An encoding device for selectively outputting encoded data of the selected encoder; 3. An encoder including a motion vector detector, and a maximum difference value among motion vector difference values for each macroblock in one picture detected by the motion vector detector is extracted. A maximum difference value extractor, wherein a difference value of a motion vector for each macroblock in one picture detected by the motion vector detector is monitored by the maximum difference value extractor, and a maximum difference in the picture is monitored. An encoding device that extracts a value and performs encoding using a minimum fcode capable of expressing the maximum difference value. 4. An encoder comprising a motion vector detector and an encoded data storage, and a maximum value among difference values of motion vectors for each macroblock in one picture detected by the motion vector detector. A maximum difference value extractor for extracting the difference value of the fd, a fcode candidate storage device for storing fcode candidates capable of expressing the maximum difference value extracted by the maximum difference value extractor, and encoding of one picture A generated bit amount counter for counting a bit amount; and a generated bit amount comparator for comparing the generated bit amount for each picture, wherein a motion vector for each macroblock in one picture detected by the motion vector detector. The maximum difference value is monitored by the maximum difference value extractor to extract the maximum difference value in the picture, and fcode candidates capable of expressing the maximum difference value are identified. Providing to the fcode candidate storage, the fc
The encoder encodes for each fcode candidate stored in the ode candidate storage, the generated bit amount counter counts the generated bit amount for each encoding, and the generated bit amount comparator generates the generated bit amount comparator. An encoding device that controls to selectively output encoded data having a minimum bit amount. 5. A method for extracting a maximum difference value among motion vector difference values for each macroblock within one picture, and using a minimum fcode capable of expressing the maximum difference value. An encoding method for encoding data.