JPH02100590A - Compensating method for stereo-dynamic picture parallax - Google Patents

Abstract

PURPOSE:To detect parallax variation through a little operation quantity, and to improve encoding efficiency by detecting the parallax variation of a background in a horizontal direction and a vertical direction by using the data of several lines portion at the upper parts of a left and a right pictures before encoding is started, and by using this parallax variation after the encoding is started. CONSTITUTION:As for a stereo video signal 101, in processing at a contact (a) side before the start of the encoding, the signal region of a part which can be considered to be background data 102 is determined by a region selection circuit 12, and it is inputted to a parallax variation detection circuit 13. Then, in the circuit 13, the parallax variations 103, 104 of a horizontal direction and a vertical direction are calculated. On the other hand, after the start of the encoding at the contact (b) side, the stereo video signal 101 is divided into the background data 105 and subject data 106 in the region selection circuit 15. The background data 105 after the start of the encoding is compensated in parallax by a parallax compensation circuit 14 by using above- mentioned parallax variations 103, 104. Consequently, since to obtain only the parallax variation of the horizontal direction is good enough for the subject 5, the parallax can be compensated through a little operation quantity.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a parallax compensation method for highly efficient encoding of stereo video images.

[Conventional technology]

In the conventional stereo moving image parallax compensation method, the motion compensation method used in high-efficiency encoding of video signals has been applied as is. By the way, it is actually difficult to precisely align two cameras, which are manually operated devices for stereo video images, in the vertical direction.

(Problem to be Solved by the Invention) Therefore, in the conventional technology, although originally it is only necessary to detect the amount of parallax displacement in the horizontal direction if only the binocular parallax is present, it is necessary to detect the displacement of the camera in the vertical direction as well. Therefore, it has become necessary to set a motion compensation range that takes vertical displacement into consideration. The camera setting positional deviation is basically unchanged during video encoding, and repeatedly detecting the amount of displacement in the vertical direction by motion compensation causes an increase in the amount of calculations and a decrease in encoding efficiency. Compensation here refers to compression means for reducing the amount of information as much as possible.

An object of the present invention is to provide a stereo video image parallax compensation method that can efficiently encode the amount of binocular parallax displacement in a stereo video image with a small amount of calculation.

(Means for Solving the Problems) The stereo video parallax compensation method according to the present invention uses data for several lines at the top of left and right images to calculate the amount of background parallax displacement in the horizontal and vertical directions before starting encoding. The direction is detected, and the amount of parallax displacement is used after the start of encoding.

(Operation) According to the present invention, the amount of parallax displacement can be detected with a small amount of calculation, and the encoding efficiency is improved.

〔Example〕

Figure 2 shows a model for photographing an object using two left and right cameras. 1 is the left camera, 2 is the left camera optical axis, 3
is the right camera, 4 is the right camera optical axis, 5 is the object (subject)
It is. In this case, disparity information must be sent in order to create a stereo image, but it is required to compress the amount of this disparity information.

Now, the vertical deviation of the two left and right cameras 1 and 3 when manually capturing a stereo video image remains about ±3 lines relative to the 240 lines of the NTSC signal even after adjustment by the naked eye. It has been known. Therefore, before starting encoding, the amount of parallax displacement of the background is detected in the horizontal and vertical directions using data for several lines at the top of the left and right images. This vertical displacement amount represents a camera setting position shift in the vertical direction, and is basically unchanged during video encoding. Therefore, even after the start of encoding, this vertical displacement can be used for vertical parallax compensation. After the encoding starts, the amount of parallax displacement may be used for background parallax compensation. In addition, to calculate the vertical displacement amount of the parallax compensation for arbitrary objects 5 located in front of the background, it is preferable to use the vertical displacement amount of the background detected before the start of encoding. For 5, it is only necessary to find the amount of horizontal parallax displacement, so it is possible to obtain a parallax correction with a small amount of calculation.

FIG. 1 is a block diagram showing one embodiment of the present invention.

In FIG. 1, reference numeral 11 denotes a changeover switch, the contact a side of which performs processing before the start of encoding, and the contact end of which performs processing after the start of encoding. 12 is a region selection circuit; 13 is a parallax displacement detection circuit;
14 is a parallax compensation circuit, 15 is a region selection circuit, 16 is a parallax displacement detection circuit, 17 is a subject parallax compensation circuit, and 18 is a region integration circuit.

Next, the operation will be explained.

The stereo video signal 101 is processed by the area selection circuit 12 . In this step, a signal region of a portion that can be considered as background data 102 is determined and manually input to the parallax displacement amount detection circuit 13. Parallax displacement detection circuit 1
3, a horizontal parallax displacement amount 103 and a vertical parallax displacement amount 104 are calculated.

On the other hand, after the start of encoding on the contact side, the area selection circuit 15
, background part data 105 and subject data 10
Divided into 6 pieces. Background part data 10 after starting encoding
5 is the horizontal parallax displacement amount 1 detected before the start of encoding.
03 and the vertical parallax displacement @ 104, the parallax is compensated in the parallax compensation circuit 14. In the parallax compensation circuit 14, a predicted image of one image of the stereo moving image to the other is created using the parallax scene, and a prediction error signal is outputted as background portion parallax compensation data +07. Subject data 1
06 is input to the parallax displacement detection circuit 16, and the horizontal parallax displacement amount 10B of the subject 5 is detected using the vertical parallax displacement N 104 of the background portion. Normally, to detect the amount of parallax displacement, one of the left and right images is shifted in an arbitrary direction and superimposed on the other, and the vector with the largest correlation value is defined as the amount of parallax displacement. Therefore, if the vertical deviation, that is, the vertical parallax displacement amount 104, is known in advance, the detection of the parallax displacement amount can be performed by adding some horizontal deviations to this vertical deviation to create a candidate vector. Then, it would be a good idea to find the vector with the highest correlation among several candidate vectors. Although the horizontal parallax displacement amount 103 varies depending on the distance m between the subject 5 and the left and right cameras 1 and 3 in FIG.
The vertical parallax displacement amount 104 due to the deviation of the setting position of No. 3 does not change. Therefore, after the start of encoding, there is no need to vary the vertical parallax displacement 104 to find the horizontal parallax displacement ft 103 that gives the maximum value of correlation, and the amount of calculation for searching for the parallax displacement amount is significantly reduced. . In the subject parallax compensation circuit 17, the amount of vertical parallax displacement 104 of the background part
Parallax compensation is performed using the horizontal direction parallax displacement amount 103 of the subject 5, and parallax compensation data 109 of the subject 5 is calculated.

The parallax compensation data 107 of the background portion and the parallax compensation data 109 of the subject 5 are combined into one in the area integration circuit 18, and a parallax compensation stereo video difference signal 110 is output.

〔Effect of the invention〕

As explained above, in this invention, before the start of encoding, the amount of parallax displacement of the background is detected in the horizontal and vertical directions using data for several lines at the top of the left and right images, and after the start of encoding, the amount of parallax displacement of the background is detected. In addition to using the amount of parallax displacement for parallax compensation,
To calculate the vertical parallax displacement amount of the parallax compensation for an arbitrary subject located in front of the background, the vertical parallax displacement amount of the background detected before the start of encoding is used, so the parallax compensation vector to be encoded and transmitted is Since it is only necessary to transmit once in the vertical direction and there is no need to repeat detecting and encoding the parallax compensation vector many times, it has the advantage of improving encoding efficiency.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.
FIG. 2 is a model diagram when photographing an object using two left and right cameras. In the figure, 1 is the left camera, 2 is the left camera optical axis, 3 is the right camera, 4 is the right camera optical axis, 5 is the object, 11 is the changeover switch, 12 is the area selection circuit, and 13 is the parallax displacement detection circuit. ,
14 is a parallax compensation circuit, 15 is a region selection circuit, 16 is a parallax displacement amount detection circuit, 17 is a subject parallax compensation circuit, 18 is a region integration circuit, 101 is a stereo video signal, 102 is background data, 103 is horizontal parallax 104 is the vertical parallax displacement amount, 105 is the background data, 106 is the subject data, 107 is the parallax compensation data, 108 is the horizontal parallax displacement amount, 109 is the parallax compensation data, and 110 is the difference between parallax compensation stereo videos. It's a signal. Figure 2

Claims (1)

[Claims] In a parallax compensation method for compressing information by removing redundancy between the left and right images of a discretized stereo video image, data for several lines at the top of the left and right images are used before the start of encoding. The amount of parallax displacement of the background is detected in the horizontal and vertical directions, and after the start of encoding, the amount of parallax displacement is used for background parallax compensation, and the amount of parallax displacement is used for the vertical direction of the parallax compensation of any subject located in front of the background. A parallax compensation method characterized in that the amount of displacement in the vertical direction of the background detected before the start of encoding is used to calculate the amount of displacement.