JPH0937301A - Stereoscopic picture conversion circuit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a stereoscopic picture conversion circuit utilizing a moving vector multiplexed to a two-dimensional (2D) video signal by MPEG algorithm. SOLUTION: A 2D video signal is supplied to a moving vector determining circuit 12, which extracts the moving vector data of each macro block from an MPEG bit stream and supplies the data to a CPU 3. The CPU 3 controls a memory control circuit 5 in accordance with the moving vector data. In addition, the CPU 3 controls a video switching circuit 6 so that the 2D video signal is switched to a left eye video signal when the direction of the moving vector is left to right and a delayed 2D video signal is applied to a right eye video signal in the case of the reverse direction.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a stereoscopic image conversion circuit for converting 2D image software into 3D image software having parallax.

[0002]

2. Description of the Related Art The applicant of the present application has proposed Japanese Patent Application No. 6-10583 as a technique for pseudo-converting a right-eye image and a left-eye image from a two-dimensional image into a three-dimensional image. .

According to this technique, a motion vector is created from a two-dimensional video image, and a relative time difference or luminance difference is generated between the right-eye video image and the left-eye video image according to the horizontal component of the created motion vector. A three-dimensional image is created by adding it.

The principle and circuit configuration will be described below with reference to FIGS.

FIG. 4 is a diagram showing the principle of the present invention, and FIGS. 5 and 6 are block diagrams of a conventional stereoscopic image conversion circuit.

First, the principle of the present invention will be described.

In a video scene in which the background does not change and the subject moves from left to right as shown in FIG. 1A, when a certain time difference is provided between the reproduced right-eye video and left-eye video as shown in FIG. 1B. The position differs by the amount of movement of the subject, and this results in parallax as shown in FIG. 7C, enabling stereoscopic viewing.

The numbers in FIGS. 1B and 1C represent field numbers.

Next, an example of the prior art will be described with reference to FIG.

In FIG. 5, one of the two-dimensional video signals from the input terminal 1 is output as it is, and the other is delayed by a predetermined field in the field memory 4 and pseudo three-dimensional video signals are obtained at the output terminals 7 and 8. . A motion vector of a part of the two-dimensional video signal is detected by the motion vector detection circuit 2, the CPU 3 controls the memory control circuit 5 according to the magnitude of the motion vector, and the field memory 4
The delay amount of is controlled.

The right-eye video signal and the left-eye video signal thus created are input to the display, and stereoscopic viewing is possible with liquid crystal shutter glasses or the like.

If a lenticular type display is used as the display, stereoscopic viewing is possible without using liquid crystal shutter glasses or the like.

In the above-mentioned conventional technique, a time difference is generated in the two-dimensional video signal as a method of providing left and right parallax. However, if the structure is such that a luminance difference is generated instead of the time difference, stereoscopic viewing is achieved by the Pulfrich effect. It is possible.

FIG. 6 shows a conventional technique using the brightness difference.
In FIG. 6, the field memory 4 and the memory control circuit 5
Instead of, the attenuator 9 and the gain control circuit 10, which attenuate between the luminance levels 0db to -10db, are controlled based on the magnitude of the motion vector. That is, when the motion of the subject is large and the motion vector is large, the brightness attenuation of the attenuator 9 is controlled to be small. When the motion of the subject is small or the motion vector is small as in slow motion reproduction, the brightness attenuation is controlled. Is controlled to increase.

[0015]

However, in the above-mentioned conventional technique, a motion detection circuit for creating and detecting a motion vector from a two-dimensional video signal is required separately.

The present invention has been made in view of the above-mentioned drawbacks, and an object of the present invention is to propose a stereoscopic video conversion circuit using a motion vector multiplexed in a two-dimensional video signal by an MPEG algorithm.

[0017]

According to the present invention, there is provided a field memory for storing a two-dimensional video signal in which motion vectors are multiplexed, and a motion vector decision circuit for detecting a motion vector in the two-dimensional video signal. CP that creates a motion amount by using the detection output from the motion vector determination circuit as an input
A stereoscopic image conversion circuit comprising U and a memory control circuit which receives a movement amount by the CPU and controls a delay amount of the field memory based on the movement amount.

Further, according to the present invention, a decoding circuit for decoding a two-dimensional video signal in which motion vectors are multiplexed, a field memory for storing the two-dimensional video signal decoded by the decoding circuit, and the two-dimensional video signal. A motion vector determination circuit for detecting the motion vector multiplexed in the above, a CPU that generates a motion amount by using the detection output from the motion vector determination circuit as an input, and a motion amount by the CPU as an input, and based on the motion amount, It is a stereoscopic image conversion circuit characterized by comprising a memory control circuit for controlling a delay amount of a field memory.

The present invention further includes an attenuator circuit for attenuating a luminance component of a two-dimensional video signal in which motion vectors are multiplexed, a motion vector determination circuit for detecting a motion vector in the two-dimensional video signal, It comprises a CPU that creates a motion amount with the detection output from the motion vector determination circuit as an input, and a level control circuit that receives the motion amount by the CPU and controls the attenuation amount of the attenuation circuit based on the motion amount. It is a featured stereoscopic image conversion circuit.

The present invention further includes a decoding circuit for decoding a two-dimensional video signal in which motion vectors are multiplexed, an attenuating circuit for attenuating the luminance component of the two-dimensional video signal decoded by the decoding circuit, A motion vector determination circuit that detects a motion vector multiplexed in a three-dimensional video signal, a CPU that creates a motion amount by using a detection output from the motion vector determination circuit as an input, and the motion amount by the CPU as an input. And a level control circuit for controlling the amount of attenuation of the attenuation circuit based on the above.

[0021]

The MPEG input terminal 11 receives a two-dimensional video signal forming a bit stream of digital data compressed by the MPEG algorithm.

One of the two-dimensional video signals is decoded by the MPEG decoder circuit 13 and supplied to the video switching circuit 6.

The other of the two-dimensional video signals is supplied to the field memory 4. The field memory 4 is variably controlled by the memory control circuit 5 on a field-by-field basis within a range from a delay amount of 0 to a maximum of 60 fields (about 1 second in the NTSC system).

The field memory output is supplied to the video switching circuit 6. The video switching circuit 6 is connected to an output terminal 7 for outputting the left-eye video signal L and an output terminal 8 for outputting the right-eye video signal R, respectively, and is controlled so that the output state is switched according to the moving direction of the subject. To be done.

The other of the two-dimensional video signal is supplied to the motion vector determination circuit 12, which extracts the motion vector data of each macroblock from the MPEG bit stream and supplies the motion vector data to the CPU 3.

The CPU 3 controls the memory control circuit 5 according to the motion vector data.

Further, the CPU 3 makes a two-dimensional video signal a left-eye video signal when the direction of the motion vector is from left to right, and makes a delayed two-dimensional video signal a right-eye video signal when the motion vector is in the opposite direction. Control 6

[0028]

EXAMPLES In recent years, DCT (Discrete Cosi)
A digital compression algorithm of a moving image using the ne Transform) has been standardized by ISO / IEC as MPEG and is being adopted for storage media, broadcasting and communication.

Two types of standardization of MPEG have been completed, MPEG1 having a bit rate of 1.5 Mbps, and
There is MPEG2, which has a bit rate of 5 Mbps or more and introduces the concept of profile and level and can widely support broadcasting and communication. For details of the compression algorithms of MPEG1 and MPEG2, see ISO / IEC, respectively.
11172 and 13818, the description thereof will be omitted here. However, the part related to the present invention is MPEG.
1 will be described below.

MPEG1 is a digital compression method based on a motion compensation interframe / intraframe adaptive predictive coding method, and a motion vector is used to perform motion compensation.
The motion vector is detected in macroblock units of 16 × 16 pixels, multiplexed with the compressed video data, and transmitted to the receiving side.

On the receiving side, interframe processing is performed based on the transmitted motion vector. However, adaptive prediction by motion compensation is not similarly performed for all video frames, and in consideration of special playback such as fast-forward / rewind by a storage medium, GOP (Group of Pi) described later is taken into consideration.
(cture) structure, forcible intraframe coding is regularly performed.

To explain the GOP structure, GO
An example of P is shown in FIG.

In MPEG1, there are three types of image types, I picture, P picture, and B picture. From I picture to the next I picture are grouped, and GO
Called P.

Next, each image type in the GOP will be described.

I pictures (I0 and I1 in the figure) are intra-frame coded, and all macroblocks are intra-frame coded. P pictures (P0 to P2 in the figure) are frames that can be forward predicted from the immediately preceding I picture or P picture, and each macroblock is subjected to intraframe / forward interframe predictive coding processing. B pictures (B0 to B7 in the figure) are frames that allow forward prediction as well as backward prediction from future frames, and each macroblock in a B picture is intraframe coded, immediately preceding I picture or Forward predictive coding from a P picture, backward predictive coding from the immediately preceding I picture or P picture, or bidirectional predictive coding processing by both forward and backward prediction is performed.

VP1, VP2, VB1F, VB1B and VB2 in the figure
F, VB3F, and VB3B indicate the relationship of motion vectors from the reference image. For example, when B2 picture is bidirectionally predicted, motion vector VB2F from I0 picture and motion vector VB2B from P1 picture are used.

In the processing of each macroblock, when predictive coding is performed between frames, a motion vector for motion compensation is multiplexed and transmitted as video data. Although an example of the GOP structure is shown here, the interval between the I picture and the P picture is variable, and the B picture may not be used depending on the profile.
In the case of MPEG2, three image types (I, P,
Although B) is the same, field-based prediction is possible, and although the image to be referred to is different from MPEG1, the present invention can be realized by a similar idea.

A feature of the present invention is that when a motion vector is already multiplexed as in MPEG, a two-dimensional video is converted into a three-dimensional video by using the motion vector.

An embodiment of the present invention will be described below with reference to FIG.

In FIG. 1, an MPEG input terminal 11 is supplied with a two-dimensional video signal forming a bit stream of digital data compressed by the MPEG algorithm.

One of the two-dimensional video signals is decoded by the MPEG decoder circuit 13 and supplied to the video switching circuit 6.

The other of the two-dimensional video signals is supplied to the field memory 4. The field memory 4 is variably controlled by the memory control circuit 5 on a field-by-field basis within a range from a delay amount of 0 to a maximum of 60 fields (about 1 second in the NTSC system). Further, the variable unit may be a small unit of one field or less.

The field memory output is supplied to the video switching circuit 6. The video switching circuit 6 is connected to an output terminal 7 for outputting the left-eye video signal L and an output terminal 8 for outputting the right-eye video signal R, respectively, and is controlled so that the output state is switched according to the moving direction of the subject. To be done.

The other of the two-dimensional video signal is supplied to the motion vector determination circuit 12, which extracts the motion vector data of each macroblock from the MPEG bit stream and supplies the motion vector data to the CPU 3.

The CPU 3 controls the memory control circuit 5 according to the motion vector data. That is, when the motion of the subject is large and the motion vector is large, the delay amount of the field memory 4 is controlled to be small, and when the motion of the subject is small or the motion vector is small such as during slow motion reproduction, the delay amount is reduced. It is controlled to be large.

The maximum number of delay fields in the field memory 4 is 60 fields, which corresponds to one second of the NTSC system, and is a time that can correspond to a normal video scene, but for slower slow motion reproduction. When used, a large capacity memory of 60 fields or more may be used. In addition, it is only necessary to delay by several hundred fields for a very slow slow motion reproduction.

Further, the CPU 3 makes the two-dimensional video signal the left-eye video signal when the direction of the motion vector is from left to right, and the delayed two-dimensional video signal the right-eye video signal when the motion vector is in the opposite direction. Control 6

Therefore, in a scene in which a subject moves horizontally in a two-dimensional video signal, parallax occurs depending on the speed of movement.

If the left and right video signals from the output terminals 7 and 8 are supplied to a lenticular type display without glasses, a stereoscopic image with a partial stereoscopic effect is pseudo even if it is a two-dimensional video signal. Can be reproduced.

Next, the operation of the motion vector determination circuit 12 will be described in detail.

The motion vector determination circuit 12 corresponds to the motion detection circuit of the prior art, but differs from the conventional circuit in that the motion detection circuit creates a motion vector from a two-dimensional video signal and is created. The motion vector for the three-dimensional image is determined from the motion vector, whereas the motion vector determination circuit 12 of the circuit of the present invention extracts the motion vector data.

Therefore, since the transmission side uses the motion vector data because it is multiplexed as data, a motion vector is created from a two-dimensional video signal like a conventional circuit, and a three-dimensional video is generated from this motion vector. The circuit for determining the motion vector of is unnecessary.

Here, in this embodiment, the motion vector determination circuit 12 has a configuration independent of the CPU 3, but it can also be adapted to software processing in the CPU.

In this embodiment, the motion vector for three-dimensionalization basically uses the motion vector from the past image, and the inverse motion vector from the future image such as the B picture is used. Motion vectors for directional prediction are not used except in special cases described later.

When a motion vector for MPEG decoding is used for three-dimensionalization, the motion vector is a macroblock unit, the I-picture is forcibly intraframe coded, and It is necessary to highly catch points where the vector frame intervals are different.

The processing method for each item will be described below.

In the MPEG algorithm, the motion vector exists in each macroblock, but in the conversion from 2D video to 3D video, it is sufficient to extract one motion vector per frame. The problem is which motion vector to extract. Therefore, in this embodiment, the motion vector is determined by the majority decision process.

Further, since the I picture has been subjected to the intra-frame coding processing, the motion vector is not multiplexed in the I picture.

Therefore, in this embodiment, when a B picture exists, the motion vector for backward predictive coding of the immediately preceding B picture is used. That is, in FIG.
The motion vector VB1B of the B1 picture is used as the motion vector of the I0 picture. At this time, the picture I0
From the perspective, it is necessary to note that the directions are reversed.

Apart from this, it is also possible to average and use the motion vectors existing in the pictures B1 and B2 before and after the picture I0.

Further, regarding the point that the frame interval of each vector is different, it is necessary to set the frame interval of the motion vector to the same value. In this embodiment, when the frame interval is unified to one frame and the motion vector VP1 (interval of 3 frames) in FIG. 3 is used, this motion vector VP1 may be set to 1/3.

In the above embodiment, the time difference is generated in the two-dimensional video signal as a method of providing the left and right parallax. However, if the luminance difference is generated instead of the time difference, the stereoscopic effect is obtained by the Pulfrich effect. Can be seen.

FIG. 2 shows this embodiment.

That is, in this embodiment, the brightness level 0d is used instead of the field memory 4 and the memory control circuit 5.
An attenuator 9 and a gain control circuit 10 that attenuate between b and -10db are used. The attenuator 9 and the gain control circuit 10 are controlled based on the magnitude of the motion vector as in the above-mentioned embodiment. That is, when the motion of the subject is large and the motion vector is large, the brightness attenuation of the attenuator 9 is controlled to be small. When the motion of the subject is small or the motion vector is small as in slow motion reproduction, the brightness attenuation is controlled. Is controlled to increase.

The attenuation amount of the attenuator 9 is 0 to 10d.
Not limited to b, when used for slow motion reproduction at a lower speed, it is advisable to give an attenuation amount of −several tens of db.

A variable gain amplifier may be used instead of the attenuator 9.

[0067]

According to the present invention, with the above-mentioned configuration, a circuit for creating a motion vector from a two-dimensional video signal,
Also, a circuit for determining a motion vector for a three-dimensional video from the created motion vector is unnecessary, and the configuration of the stereoscopic video conversion circuit can be simplified.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a stereoscopic image conversion circuit of the present invention.

FIG. 2 is a block diagram showing another embodiment of the stereoscopic image conversion circuit of the present invention.

FIG. 3 is a diagram showing a format of a two-dimensional video signal input to the stereoscopic video conversion circuit of the present invention.

FIG. 4 is a diagram illustrating a principle for creating a three-dimensional image.

FIG. 5 is a block diagram of a conventional stereoscopic video conversion circuit.

FIG. 6 is a block diagram of a conventional stereoscopic image conversion circuit.

[Explanation of symbols]

 3 CPU 4 field memory 5 memory control circuit 6 video switching circuit 7 output terminal 8 output terminal 11 input terminal 12 motion determination circuit 13 MPEG decoder circuit

Claims (5)

[Claims] 1. A field memory for storing a two-dimensional video signal in which motion vectors are multiplexed, a motion vector determination circuit for detecting a motion vector multiplexed in the two-dimensional video signal, and detection from the motion vector determination circuit. A stereoscopic image conversion circuit, comprising: a CPU that produces an amount of motion with an output as an input; and a memory control circuit that receives the amount of motion by the CPU as an input and controls a delay amount of the field memory based on the amount of motion. . 2. A decoding circuit for decoding a two-dimensional video signal in which motion vectors are multiplexed, a field memory for storing the two-dimensional video signal decoded by the decoding circuit, and a multiplexing circuit for the two-dimensional video signal. A motion vector determination circuit that detects a motion vector, a CPU that generates a motion amount by using a detection output from the motion vector determination circuit as an input, and a motion amount by the CPU as an input, and a delay of the field memory based on the motion amount. A stereoscopic image conversion circuit comprising a memory control circuit for controlling the amount. 3. An attenuation circuit for attenuating a luminance component of a two-dimensional video signal in which motion vectors are multiplexed, a motion vector determination circuit for detecting a motion vector multiplexed in the two-dimensional video signal, and the motion vector determination circuit. And a level control circuit for inputting the movement amount by the CPU and controlling the attenuation amount of the attenuation circuit based on the movement amount. Video conversion circuit. 4. A decoding circuit for decoding a two-dimensional video signal in which motion vectors are multiplexed, and an attenuating circuit for attenuating the luminance component of the two-dimensional video signal decoded by the decoding circuit.
A motion vector determination circuit for detecting a motion vector multiplexed in the two-dimensional video signal, and a CPU for generating a motion amount by using a detection output from the motion vector determination circuit as an input,
A stereoscopic image conversion circuit comprising: a level control circuit which receives a movement amount by the CPU as an input and controls the attenuation amount of the attenuating circuit based on the movement amount. 5. The motion vector multiplexed in the two-dimensional video signal is a motion vector digitally compressed by an MPEG algorithm.
The stereoscopic image conversion circuit according to any one of claims 1 to 4.