JPH10257495A - Image data compression encode method and image data compression encoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide satisfactory picture quality or compression by encoding the structure type of a B picture positioned between a rear reference picture and a fore reference picture based on the motion vector data for detecting it as the same structure type as the structure type of the rear reference picture. SOLUTION: A selector 5 selects either fore image data (b) read out of a fore reference image memory 6 or rear reference image data (c) read out of a rear reference image memory 7. A motion vector detection circuit 2 finds motion vector data (f) by inputting input image data (a) and the reference image data selected by the selector 5. The motion vector detection circuit 2 finds a difference (i) between the input image and the reference image at the time of finding the motion vector data (f), compares the total sum of difference (h) found by a motion vector detection circuit 1 with the total sum of difference (i), the structure type reducing the total sum is selected and its flag data (k) are sent out.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data compression encoding method and an image data compression encoder used for image data compression.
The present invention relates to an image data compression encoding method and an image data compression encoder used for image data compression based on H.2.

[0002]

2. Description of the Related Art In an image data compression encoder based on MPEG2, a frame structure and a field structure can be selected for each picture. For example, as shown in FIGS. 4 and 5, in an image data compression encoder that automatically adapts to both frame / field structure types, the forward motion vector detecting circuits 11 and 12 and the backward motion vector detecting circuits 11 and 12 are provided for each structure type. , A total of four motion vector detecting circuits 13 and 14 are used. In this specification, a frame structure and a field structure are described as a structure type type.

In the image data compression encoder shown in FIGS. 4 and 5, reference numerals 16 and 17 denote motion vector detection circuits 11 to 11 between the image data as input image information.
Reference numeral 14 denotes a forward reference image memory and a backward reference image memory that respectively store forward reference image data and backward reference image data for detecting motion vector data.

[0004] The image data compression encoder shown in FIG.
The frame structure encoding circuit 15 receives the forward motion vector data of the frame structure detected by the motion vector detecting circuits 11 and 13 and the backward motion vector data of the frame structure, encodes the frame structure, and detects by the motion vector detecting circuits 12 and 14. The field structure encoding circuit 18 receives the forward motion vector data of the field structure and the backward motion vector data of the field structure, encodes the field structure, and outputs the bit streams output from the encoding circuits 15 and 18 to the bit stream comparator 19. The better comparison is selected, that is, the smaller number of generated bits is selected and output as a bit stream.

In the image data compression encoder shown in FIG. 5, motion vector data is obtained by the motion vector detection circuits 11 to 14, and at the same time, difference data between an input image and a reference image obtained by using the obtained motion vector data is obtained. The difference sum comparator 20 for calculating the sum of one picture of the difference data sent from the motion vector detection circuits 11 to 14 and comparing the sums determines the structure type in which the sum becomes smaller, Sum comparator 2
0 by the selectors 21 and 22 based on the structure type determined by the motion vector detection circuit 1
1 and 12 and one of the motion vector data output from the motion vector detection circuits 13 and 14 is selected, that is, the frame structure motion vector data or the field structure motion vector data. And the encoding circuit 23 receiving the selected motion vector data
And output as a bit stream.

As described above, in the image data compression encoder shown in FIG. 4, encoding is actually performed using the frame structure and the field structure, and the result can be compared, that is, the one with a smaller number of generated bits. Is selected. In the image data compression encoder shown in FIG.
At the same time as obtaining the motion vector data in step 4, the sum of one picture of the difference data between the input image and the reference image when the obtained motion vector data is used is obtained, and the structure type having the smaller sum is selected. I have. Generally, the configuration of FIG. 5 is considered to have a smaller circuit scale.

[0007]

However, according to the above-mentioned conventional image data compression encoder, there is a problem that the circuit scale becomes large because a large number of motion vector detection circuits are required. Further, by fixing the structure type to the field structure or the frame structure, the number of motion vector detecting circuits can be reduced to two, but there is a problem that it is difficult to obtain a good image quality and a good compression ratio.

An object of the present invention is to provide an image data compression encoding method and an image data compression encoder capable of reducing the number of motion vector detection circuits to two and obtaining good image quality and compression ratio.

[0009]

According to the present invention, the structure type of a B picture located between a backward reference picture and a forward reference picture is the same as the structure type of the backward reference picture. The encoding is performed based on the motion vector data detected as.

According to the image data compression encoding method of the present invention, the structure type of the B picture located between the backward reference picture and the forward reference picture is the same as the structure type of the backward reference picture. Encoding is performed based on the detected motion vector data. Accordingly, only two motion vector detecting means are required for obtaining motion vector data, and the same image quality and compression ratio as those obtained when four motion vector detecting means are obtained can be obtained. Is not reduced.

An image data compression encoder according to the present invention comprises a forward reference image memory, a backward reference image memory, forward reference image data read from the forward reference image memory, and a backward reference image read from the backward reference image memory. Selecting means for selecting one of the image data, first motion vector detecting means for obtaining motion vector data from the input image data and the forward reference image data, and reference image data output from the selecting image data. A second motion vector detecting means for obtaining motion vector data, a sequencer for designating a picture type to be sequentially processed with respect to an input image and switching a selecting means based on the designated picture type, and a P picture or an I picture by the sequencer. Output from the first motion vector detecting means when instructed. Of the difference data between the input image and the reference image and the difference data between the input image and the reference image output from the second motion vector detecting means, and determines the structure type as the structure type in which the sum of one picture is smaller. And when the P picture or the I picture is indicated by the sequencer, the first and second motion vector detecting means obtain the forward motion vector data of the frame structure and the forward motion vector data of the field structure. When the B picture is indicated by the sequencer and when the difference sum comparator determines the frame type as the frame type, the first and second motion vector detecting means detects the forward motion vector of the frame structure. The first and second motions are obtained when a B picture is indicated by the sequencer and the field type is determined as the structure type by the difference sum comparator when the backward motion vector data of the frame structure is obtained. The forward motion vector data of the field structure and the backward motion vector data of the field structure are obtained by the vector detecting means, and the motion vector data outputted from the first and second motion vector detecting means and the picture type information designated by the sequencer. And encoding based on the following.

According to the image data compression encoder according to the present invention, the motion vector detection requires only two motion vector detecting means, the first motion vector detecting means and the second motion vector detecting means. The same image quality and compression ratio can be obtained, so that good image quality can be obtained and the compression ratio does not decrease.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An image data compression encoder according to the present invention will be described below with reference to an embodiment. FIG. 1 is a block diagram showing a configuration of an image data compression encoder according to one embodiment of the present invention.

In the image data compression encoder E according to one embodiment of the present invention, a forward reference image memory 6 storing image data of a forward reference image corresponding to an input image based on input image data a; The backward reference image memory 7 in which image data of the backward reference image corresponding to the input image based on the data a is stored, and the input image data and the forward reference image data read out from the forward reference image memory 6 are input and moved. A motion vector detection circuit 1 is provided for obtaining difference data h between an input image and a reference image when the motion vector data d obtained is used at the same time as obtaining the vector data d.

Further, in the image data compression encoder E, a selector 5 for selecting one of the forward image data b read from the forward reference memory 6 and the backward reference image data c read from the backward reference image memory 7. And the input image data a and the reference image data selected by the selector 5 are input to obtain the motion vector data f, and the difference data between the input image and the reference image when the obtained motion vector data f is used. A motion vector detecting circuit 2 for calculating i, a sum of one picture of the difference data h and i is calculated, and the sum is compared to determine a structure type as a structure type having a smaller sum. Hold until the structure type is determined, and determine the flag based on the determined and held structure type. Motion data k and a difference sum comparator 3 to deliver respectively to the vector detection circuit 1 and the encoding circuit 4 which will be described later.

Further, in the image data compression encoder E, buffer memories 9 and 10 for storing the motion vector data d and f obtained by the motion vector detection circuits 1 and 2 respectively, and the motions read from the buffer memories 9 and 10 Vector data e, g, flag data k, 1
The input image data a, the forward reference image data b, and the backward reference image data c, each of which is delayed by one picture by the picture delay circuits 8B, 8C, and 8D, are supplied and encoded, and the I picture, P
A sequencer 8 for outputting an instruction signal m for pictures and B pictures, and switches the selector 5 based on the instruction signal m.

The input image data a, forward reference image data b, and backward reference image data c input to the encoding circuit 4 are data delayed by one picture from the input image corresponding to the input image data a. This is because the motion vector data read from the buffer memories 9 and 10 is delayed by one picture.

Next, before describing the operation of the image data compression encoder E, the basic principle of the image data compression encoder E will be described with reference to FIGS.

In FIG. 2, A, B, C, and D are input images input to the image data compression encoder. The input images A, B, C, and D are schematically shown. , C, d are entered in this order.

In an MPEG2 image data compression encoder, a P picture uses only a forward reference picture motion vector based on a forward reference picture. For example, in FIG. 2, a P picture (D) is a P picture (A) in front of it.
Is a forward reference image, and only the forward motion vector (e) from now on is used. Therefore, if two motion vector detection circuits 1 and 2 are provided, it is possible to obtain forward motion vector data corresponding to each of the case where the frame structure is adopted for the P picture and the case where the field structure is adopted for the P picture.

In encoding a B picture, motion vectors (a forward reference motion vector and a backward reference motion vector) from two forward and backward reference pictures are used. For example, a B picture (a) in FIG. 2 uses a P picture (a) as a forward reference picture and a P picture (d)
Is a backward reference image, and encoded using the forward reference motion vector (f) and the backward reference motion vector (g).
If it is attempted to obtain a motion vector corresponding to the frame structure and a motion vector corresponding to the field structure for the forward reference motion vector (f) and the backward reference motion vector (g), four motion vector detection circuits are required. Would.

The B picture (C) in FIG. 2 uses the P picture (A) as the forward reference picture, the P picture (D) as the backward reference picture, and the forward reference motion vector (H) and the backward reference motion vector (K). Is encoded using If it is attempted to obtain a motion vector corresponding to the frame structure and a motion vector corresponding to the field structure for the forward reference motion vector (h) and the backward reference motion vector (h), four motion vector detection circuits are required. This is the same as in the case of the B picture (a).

In MPEG2 image data compression encoding, the order of the original image shown in FIG.
As shown in (b), the order of the encoding process is changed, and the order of the images is changed. In the image data compression encoder E according to the embodiment of the present invention, the structure type of the B picture is determined as follows.

A P picture (D) is a B picture (A) or a B picture (B).
Because it is encoded before the picture (c),
The structure type adopted in P picture (D) is B
The same applies to picture (a) and B picture (c).
That is, assuming that a frame structure is selected as the structure type of the P picture (d), the B picture (b) and the B picture (c) are also frame structures, and the two motion vector detection circuits 1 and 2 correspond to the frame structure type. The obtained forward reference motion vector and backward reference motion vector are obtained and output.

In general, which of the frame structure and the field structure should be used as the structure type depends on the speed of movement of the image.
Therefore, assuming that the field structure is adopted in the P picture (D), the movement from the P picture (A) to the P picture (D) is fast, that is, the change amount of the image is large in many cases.

Therefore, at this time, B pictures (A) and B pictures (B) and P pictures (P) (P)
Since it is estimated that the amount of change of the image is also large in picture (c), it is possible that better results can be obtained by adopting a field structure for these two B pictures (a) and B picture (c). It is considered high. That is, by configuring the structure type of the B picture to adopt the same structure type as that of the P picture immediately following the B picture, the motion vector detecting circuit can be configured to have 4
Even without the full field / frame adaptation used, very close results can be obtained.

As described above, in the image data compression encoder E, the structure type of the B picture is determined in advance to either the frame structure or the field structure, so that only two motion vector detecting circuits are required. is there.

In the above example, a B picture between a P picture and a P picture is considered (in FIG. 3A). However, even if the preceding P picture is changed to an I picture,
For example, the same is true even if the picture of the P picture (A) in FIG. 2 is an I picture. On the other hand, when the subsequent P picture changes to an I picture, for example, FIG.
In the case where the picture (d) becomes an I picture, the motion vector can be detected assuming that the picture (d) is a P picture and a structure can be selected.

Next, the operation of the image data compression encoder E will be described. The input image data a based on the input image according to the order shown in FIG. 3B is sequentially supplied, and the forward reference image data b based on the forward reference image corresponding to the input image is stored in the forward image memory 6. The backward reference image data c based on the backward reference image corresponding to the image is stored in the backward image memory 6. Specifically, FIG.
In the case of (b) as an example, when an image of a picture B1 or a picture B2 is input, the forward reference memory 6
Stores image data b based on the image of the picture I0, and the backward reference memory 7 stores image data c based on the image of the picture P3.

When the input image is an image to be processed as a P picture or an I picture, the sequencer 8 outputs a high-potential signal as an instruction signal m, and outputs an identification signal for identifying the P picture and the I picture. When the input image is an image to be processed as a B picture, a low-potential signal is output as the instruction signal m.
The instruction signal m and the identification signal are delayed by one picture by the one-picture delay circuit 8A and supplied to the encoding circuit 4 as an instruction signal m '. This is because the encoding circuit 4 operates at a timing delayed by one picture.

First, the operation when a P picture is designated by the instruction signal m from the sequencer 8 will be described. Since only the forward reference image is used in the P picture, the selector 5 receiving the instruction signal m outputs the image data b read from the forward reference image memory 6 to the motion vector detection circuit 2. Therefore, both the motion vector detection circuits 1 and 2 obtain a motion vector from the input image data a and the forward reference image data b.

At this time, the motion vector detection circuit 1 outputs the motion vector data d corresponding to the frame structure, and the motion vector detection circuit 2 outputs the motion vector data f corresponding to the field structure. Contrary to the motion vector detection circuits 1 and 2, the motion vector detection circuit 1 outputs motion vector data corresponding to the field structure,
After that, motion vector data corresponding to the frame structure may be output.

The motion vector data d output from the motion vector detection circuit 1 is stored in a buffer memory 9, and the motion vector data f output from the motion vector detection circuit 2 is stored in a buffer memory 10. The buffer memories 9 and 10 can store motion vector data for one picture. Further, the motion vector detection circuits 1 and 2 output difference data h and i between the input image and the reference image when the output motion vector data is used, to the difference sum comparator 3.

The difference sum comparator 3 calculates the sum of the difference data for one picture for each of the difference data h and i, selects the structure type having the smaller value, and outputs it as the structure type signal k. . The structure type signal k is held until the structure type of the next P picture is determined.

The encoding circuit 4 reads out a motion vector from the buffer memory 9 or 10 based on the structure type signal k and performs encoding. That is, when the structure type signal k indicates a field structure, a motion vector is read from the buffer memory 9, and when the structure type signal k indicates a frame structure, a motion vector is read from the buffer memory 10. Encoding is performed.

The input image data a
And the reference image data b and c are input so as to be synchronized with the motion vector data, and are encoded in cooperation with the motion vector data, and the encoded bit stream j is output.

Next, the operation when a B picture is designated by the instruction signal m will be described. When a B picture is designated, the backward reference image data c read from the backward reference image memory 7 by the selector 5 is sent to the motion vector detection circuit 12. This is because a B picture uses a forward reference image and a backward reference image,
The selector 5 outputs the image data c read from the backward reference image memory 7 to the motion vector detection circuit 12.

The motion vector detecting circuit 1 obtains the forward motion vector data d, and the motion vector detecting circuit 2 obtains the backward motion vector data f. Each of the motion vector detection circuits 1 and 2 calculates and outputs motion vector data corresponding to the structure type indicated by the structure type signal k. The motion vector data obtained in this manner is stored in the buffer memories 9 and 10 and then read out and read out according to an instruction from the encoding circuit 4 and used for encoding.

When an I picture is designated by the instruction signal m, the operation is the same as that for a P picture except that the encoding circuit 4 does not use motion vector data.

As described above, the image data compression encoder E includes two motion vector detection circuits, ie, motion vector detection circuits 1 and 2, and the motion vector detection circuits 1 and 2 perform the following combination modes α and Operated at β, γ.

When detecting a motion vector for a P picture, the motion vector detection circuits 1 and 2 operate in a mode α for obtaining forward motion vector data of a frame structure and forward motion vector data of a field structure. The forward motion vector data for each structure type of the / field structure is obtained, and at the same time, the difference data between the input image and the reference image when the motion vector data is used is obtained and output.

The motion vector detection circuits 1 and 2 are for detecting a motion vector for a B picture,
When instructed to use the frame structure, it operates in the mode β for obtaining the forward motion vector data of the frame structure and the backward motion vector data of the frame structure, and when instructed to use the field structure, It operates in a mode γ for obtaining a forward motion vector of a structure and a backward motion vector of a field structure, and outputs forward and backward motion vector data.

As the structure type of the P picture, a structure type in which the total sum of the difference between the input image and the reference image based on the input image data a is smaller is selected, and the structure type of the B picture is the input order (FIG. 3).
In (a)), the structure type of the P picture immediately following the B picture (the P picture processed immediately before the B picture in the processing order (FIG. 3B)) is selected.

As described above, in the image data compression encoder E, only two motion vector detecting circuits are required.

In the image data compression encoder E according to the embodiment of the present invention, the case where the structure type of the P picture is determined by comparing the sum of the difference picture between the input picture and the reference picture is exemplified. However, the encoding may be performed by actually encoding with each structure type and comparing the generation amount of the output bit stream.

[0046]

As described above, according to the image data compression encoding method and the image data compression encoder according to the present invention, good results can be obtained in the compression ratio and the image quality without increasing the circuit scale.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of an image data compression encoder according to an embodiment of the present invention.

FIG. 2 is a schematic diagram for explaining an encoding operation performed by the image data compression encoder according to the embodiment of the present invention;

FIG. 3 is a schematic diagram for explaining an operation of encoding by the image data compression encoder according to the embodiment of the present invention;

FIG. 4 is a block diagram showing a configuration of a conventional image data compression encoder.

FIG. 5 is a block diagram showing another configuration of a conventional image data compression encoder.

[Explanation of symbols]

 1 and 2 motion vector detection circuit 3 difference sum comparator 4 encoding circuit 5 selector 6 forward reference image memory 7 backward reference image memory 8 sequencer 9 and 10 buffer memory E image data compression encoder

Claims (2)

[Claims] 1. Encoding a structure type of a B picture located between a backward reference picture and a forward reference picture based on motion vector data detected as being the same as the structure type of the backward reference picture. An image data compression encoding method characterized by the following. 2. A method for selecting one of a forward reference image memory, a backward reference image memory, and forward reference image data read from the forward reference image memory and backward reference image data read from the backward reference image memory. Selecting means, first motion vector detecting means for obtaining motion vector data from input image data and forward reference image data, and second motion for obtaining motion vector data from input image data and reference image data output from the selecting means. Vector detecting means, a sequencer for designating a picture type to be sequentially processed with respect to an input image and switching a selecting means based on the designated picture type, a first motion when a P picture or an I picture is designated by the sequencer Difference data between the input image output from the vector detection means and the reference image, (2) a difference sum comparator for determining a structure type for a structure type in which the sum of the difference data between the input image output from the motion vector detecting means and the reference image is smaller by one picture, and a difference sum comparator for determining the structure type; When a picture or an I-picture is indicated When the first and second motion vector detecting means determine the forward motion vector data of the frame structure and also when the B-picture is indicated by the sequencer When the structure type is determined to be the frame structure by the difference sum comparator, the first and second motion vector detection means obtain forward motion vector data of the frame structure and perform frame structure. When the B picture is designated by the sequencer and the structure is determined to be the field type as the structure type by the difference sum comparator, the field structure is determined by the first and second motion vector detecting means. And the backward motion vector data of the field structure are obtained, and the encoding is performed based on the motion vector data output from the first and second motion vector detecting means and the picture type information specified by the sequencer. An image data compression encoder characterized by the above-mentioned.