JPH03200007A - Stereoscopic measuring instrument - Google Patents

Abstract

PURPOSE:To easily allow plural pictures to correspond to each other and to simplify a measuring system by executing picture conversion processing for expanding or contracting a picture by using the ratio of the focal distances of lenses in right and left image pickup devices as its magnification. CONSTITUTION:The image pickup devices 101, 102 output pictures obtained by picking up an subject image. The pictures obtained by the devices 101, 102 are respectively set up as pictures L, R and the picture R is outputted to a correspondence processing part 104. When the focal distances of the lenses in the devices 101, 102 are respectively defined as fL, fR (fL<fR), a picture conversion processing part 103 executes the picture conversion processing for expanding or contracting the picture L outputted from the device 101 by using the ratio fR/fL of the focal distances of the lenses in the devices 101, 102 as the magnification and outputs a picture L' obtained by the picture conversion processing of the screen L to the processing part 104. The processing part 104 allows same picture elements in a real space to correspond to each other out of picture elements in the two pictures L', R and outputs the coordinate values of respective corresponding picture elements in the pictures L', R to a depth information extracting part 106.

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to a stereo measurement device that extracts depth information, which is the distance from a camera to each part of a subject, using stereo images.

<Conventional technology> Conventional stereo measurement devices are obtained by photographing the same scene with two cameras installed in different locations, or by photographing the same scene twice with one camera in different locations. Correlate things that are the same in real space on two screens,
It extracts depth information based on the principle of triangulation.

<Problems to be Solved by the Invention> In conventional measuring devices, two cameras must be placed sufficiently apart in order to extract depth information correctly without being affected by image quantization errors. Therefore, 11! There are problems in that the measurement system becomes large-scale, and the screens obtained by taking pictures with two cameras are greatly different, so that incorrect processing may occur in the correspondence between the screens, or it may not be possible to make the correspondence.

SUMMARY OF THE INVENTION In order to solve these conventional problems, it is an object of the present invention to provide a stereo measurement device that allows easy correspondence between screens and has a simple observation system.

Means for Solving the Problems> The configuration of the present invention to solve the above problems is as follows: The lenses used have different focal lengths, and two cameras are used to simultaneously photograph the same subject and output a screen containing the subject image. image pickup devices arranged side by side, and one of the two screens output from these two image pickup devices is subjected to screen conversion processing that enlarges or reduces the ratio of the focal lengths of the lenses of both image pickup devices as a magnification factor. , a screen conversion processing unit that outputs a converted screen that has undergone screen conversion processing, and a screen conversion processing unit that outputs a converted screen that has undergone screen conversion processing, and a screen conversion processing unit that outputs a converted screen that has undergone screen conversion processing, and a screen conversion processing unit that connects each pixel of the conversion processed screen and each pixel of the other screen that has not undergone screen conversion processing, so that they are the same in real space. The association is characterized by having a processing unit and a depth information extraction unit that extracts depth information that is the distance from the imaging device to each part of the subject based on the coordinates of each associated pixel.

In the present invention, lenses with different focal lengths are used in the left and right imaging devices to simultaneously capture two left and right images of the same scene, and the left and right images are displayed on one of the two captured images. Screen conversion processing is performed to enlarge or reduce the ratio of the focal length of the device's lens as a magnification, and depth information is obtained from two screens: the other screen that has not been subjected to screen conversion processing, and the conversion processing screen that has undergone screen conversion processing. Extract.

<Embodiments to be Produced> Hereinafter, embodiments of the present invention will be described in detail based on the drawings. FIG. 1 shows a configuration diagram of an embodiment of the present invention in the case where a screen conversion process is performed on the screen obtained from the left imaging device. In the figure, 101 is an imaging device on the left side, 102 is an imaging device on the right side, 103 is a screen conversion processing unit, 104 is a processing unit for mapping, 105 is an observation system parameter storage unit, 1
06 is a depth information extraction unit.

The imaging device 101 and the imaging device 102 are arranged side by side so that the optical axes of both lenses are parallel, and photograph the same subject (scene) at the same time. In this case, the imaging devices 101 and 102 use different focal lengths of lenses. The imaging devices 101 and 102 then output a screen on which the subject image is projected.

The screen captured by the imaging device 101 is used as a screen image, and this screen image is output to the screen conversion processing unit 103. Imaging device 10
2 is taken as screen R, and this screen R is output to the processing unit 104 for correspondence. In addition, the imaging device 101゜1
The focal length of the lens at 02 is fL, f
Let R(fL<f,).

The screen conversion processing unit 103 performs screen conversion processing on the screen output from the imaging device 101 to enlarge or reduce it using a magnification of f, /fL, which is the ratio of the focal lengths of the lenses of the imaging devices 101 and 102. The screen L' obtained by performing the screen conversion process is output to the mapping processing section 104.

Screen conversion processing methods for enlarging or reducing the screen include scaling methods used in intelligent copies with editing functions, such as the SPC method, the logical sum method,
Various methods such as a projection method, a 9-division method, and a high-speed projection method can be adopted.

The processing unit 104 determines that each pixel on the two screens, the screen L' output from the screen conversion processing unit 103 and the screen R output from the right imaging device 102, is the same in real space. The depth information extraction unit 106 associates the coordinate values of each pixel on the screen L' and the screen R that are associated with each other.
Output to. The process of mapping between two screens is, for example,
“Computer Vision” Yoshiaki Shirai Shokodo pp, 65
-71J, the processing unit 104 can realize the correspondence.

Here, the method of this correspondence processing will be outlined with reference to FIG. For example, to find the point P8 on the right screen that corresponds to the point PL on the left screen, first, a window WL containing the point PL in the center is set. Next, in the right screen, decide on multiple candidates to be the corresponding point of point PL, compare each window with each candidate point as the center, and find the one in the window that is closest to the window point. Let the point be the corresponding point P6. In the figure, the images of swear and W8 are similar, but - and W
'8 are different, so it can be seen that point P'8 in W'8 is not a corresponding point. The similarity between windows is determined by evaluating the brightness of the windows and the variance of the brightness.

On the other hand, the parameter storage unit 105 for the U side yarn in FIG.
is the focal length g of the lens in the imaging devices 101 and 102
lfL and fFl, and the imaging device 1 that was actually measured in advance
The distance 1IIZFIL between the imaging planes 01 and 102 is stored, and the stored values of fL, fR, and ZRL are output to the depth information extraction unit 106.

The depth information extraction unit 106 associates the coordinate values in the screen L′ and screen R of the same point in the real space output from the processing unit 104 with the observation system output from the parameter storage unit 105 of the U side thread. Extract depth information from parameters.

An example of the depth information extraction unit 106 will be described below.

The coordinate system in this example is shown in FIG.

If the positions of the left and right imaging devices 1101 and 102 are different from those shown in FIG. Can you find the screen obtained when the left and right imaging devices 101 and 102 are arranged as shown in the figure? When the positions of the left and right imaging devices 101 and 102 are as shown in FIG. 2, the position of the point P in the scene in real space and the screen image.

The positional relationship on the screen R is as shown in FIG.

Therefore, the coordinate value of point P in real space is p (xp yp
Z) Coordinate values of N point P on the screen (UL, VL)
, let the coordinate values on screen R be (UFl, V,),
The X and Y coordinate values of the spatial position of point P can be expressed by the following equations.

Regarding screen size, Y8L is the distance between the optical axes of the lenses in the left and right imaging devices 101 and 102, and z8. is the distance between the imaging surfaces of the left and right imaging devices.

Eliminating X from equations (1) and (3) yields.

In addition, the screen conversion processing unit 103 performs screen conversion processing on the screen output from the right imaging device 101 using f, /fL, which is the ratio of the focal lengths of the lenses in the left and right imaging devices, as a magnification. , on the screen (UL, v
L) and the coordinate values (U', V'L) of that point on the screen L' after image conversion, the following holds for the screen R. Substituting equation (6) into equation (5) yields. Depth information of point P is obtained from equation (7).

That is, the depth information extraction unit 106 determines the coordinate values of each pixel on the screen L' and the screen R that are associated with each other in the real space outputted from the processing unit 104 as shown in FIG. The X components U', , UFl and the paran-1 of the observation system output from the parameter storage unit 105 shown in FIG.
Depth information can be extracted by calculating the Z coordinate value of each point P (x, y, z) from equation (7).

Furthermore, the correspondence is output from the processing unit 104. Among the coordinate values, the coordinate values on the screen R of each point P in the real space (horse,
R8), the parameters of the observation system output from the parameter storage unit 105, and the depth information of point P calculated from equation (7), the spatial position of point P is determined from equations (3), 2, and (4). You can also ask for .

<Effects of the Invention> As explained above, according to the present invention, lenses with different focal lengths are used in the left and right imaging devices, so that one of the two left and right images captured for the same scene can be , a screen conversion process is performed to enlarge or reduce the ratio of the focal lengths of the lenses of the left and right imaging devices as a magnification, and the other screen is not subjected to the screen conversion process, and the conversion process screen is subjected to the screen conversion process. Since depth information can be extracted from the screen, the left and right imaging devices can be placed adjacent to each other, making it possible to realize a stereo measurement device with a simple observation system and less depth information extraction errors caused by incorrect processing of association.

In other words, since the stereo measurement device of the present invention can extract depth information regardless of the distance between the left and right imaging devices,
The left and right imaging devices can be placed adjacent to each other.

When the left and right imaging devices are placed adjacent to each other, the similarity between the two screens output from the screen converting means is high, making it easy to correlate the screens and simplifying the observation system.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram showing an embodiment of the present invention, FIG. 2 is an explanatory diagram showing a coordinate system in the embodiment, FIG. 3 is an explanatory diagram showing the relationship between spatial position and position on the screen, and FIG. The figure is a diagram showing the principle of the correspondence method. In the drawing 0 0 0 0 0 0, 1 is the left imaging device, 2 is the right imaging device, 3 is the screen conversion processing unit, 4 is the processing unit for correspondence, 5 is the observation system parameter storage unit, 6 is the depth This is the information extraction part. Figure 4 Patent applicant: Agent of Nippon Telegraph and Telephone Corporation

Claims (1)

[Claims] Two imaging devices juxtaposed, whose lenses have different focal lengths and which simultaneously photograph the same subject and output a screen containing the subject image; A screen conversion process that performs screen conversion processing to enlarge or reduce one of the two output screens using the ratio of the focal lengths of the lenses of both imaging devices as a magnification, and outputs the converted screen after the screen conversion process. a correspondence processing unit that associates each pixel of the conversion processing screen with each pixel of the other screen that has not undergone the screen conversion process so that they are the same in real space; A stereo measurement device comprising: a depth information extraction unit that extracts depth information that is a distance from an imaging device to each part of a subject based on coordinates;