JP2629926B2 - Range image acquisition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a range image acquisition device used for a three-dimensional measurement device.

[Conventional technology]

As a device for obtaining a three-dimensional position of an object or the like from an image input from a television camera, a range image acquisition device has been proposed (Japanese Patent Application Laid-Open No. 62-195379). An example of a method for obtaining the distance from the camera to the object by the distance image acquisition device will be briefly described with reference to FIG.

The image data acquired by the two sensors 5 and 6 having different spectral sensitivities σ ₁ and σ ₂ of the camera 3 are passed through A / D converters 7 and 8 for each pixel of the two image memories 9 and 10. The ratio R is calculated according to the following equation, and the ratio image is stored in the ratio image memory 12.

Here, I ₁ : the output of the pixel of the sensor 5 where the point P is imaged. I ₂ : the output of the pixel of the sensor 6 where the point P is imaged. When a white plate is placed on the plane 19 and the first calibration image is obtained, The function operation unit 14 is connected to the ratio image memory via the switch 13.
A function X _a (R) of the ratio R on the plane 19 is obtained from the ratio image stored in 12 and is held. The function X _a (R) has a value in the range of observed R _a1 ≦ R ≦ R _a2 . Similarly, when a white plate is placed on the plane 20 and the second comparative image is acquired, the function operation unit 15 outputs the ratio R in the plane 20 from the ratio image stored in the ratio image memory 12 via the switch 13. Function X _b (R)
And hold. The function X _b (R) has a value in the range of R _b1 ≦ R ≦ R _b2 .

The diffraction position calibration unit 16 uses the appropriate constant c for the functions X _a (R) and X _b (R) according to Equation (2) for R in the overlapping part of the domain of both Rs, using the appropriate constant c. _a , Z _a ), (X
_b , Z _b ) is obtained, and the diffraction position (X ₀ , Z ₀ ) is obtained from Expressions (3) and (4). Calibrated diffraction position (X ₀ , Z ₀ )
Are held in the diffraction position calibration unit 16.

Z = a _R X + b _R (R = R ₁ , R ₂ ,...) (3) When measuring the distance to the actual object 2, the actual scene is obtained in the same way in the image memories 9 and 10, and the ratio image is obtained in the ratio image memory 12. The distance calculation unit 17 reads out the ratio R together with the X coordinate value from the ratio image memory 12 via the switch 13 for each image, and obtains the diffraction position (X ₀ , Z ₀ ) held in the diffraction position calibration unit 16 and the function. Function X stored in operations 14 and 15
_The distance Z is obtained from the equation (5) or (6) using _a (R) and _Xb (R), and is written into the distance image memory 18.

As a result of the above series of operations, the distance Z from the camera 3 to the object 2 is obtained for each pixel of the distance image memory 18.

[Problems to be Solved by the Invention] According to the above-described related art, two or more measurements are required to reduce a region where distance measurement becomes impossible due to a blind spot of a spectral pattern. For example, in FIG. 9, the area A is visible from the camera, and can be measured because the spectral pattern is projected. But region B
Is visible from the camera, but cannot be measured because no spectral pattern is projected. In order to measure both the regions A and B, it is necessary to rotate the object or rotate the camera and the projection device to perform the measurement twice or more, so that it takes time.
This is a problem particularly when measuring an object such as a human body, which is difficult to maintain a constant posture for a long time.

SUMMARY OF THE INVENTION An object of the present invention is to provide a range image acquisition apparatus capable of reducing an unmeasurable region caused by a blind spot of a projected spectral pattern with only one measurement without performing mechanical scanning.

[Means for solving the problem]

The distance image acquisition apparatus according to the present invention includes a plurality of spectral pattern projecting means for projecting a spectral pattern in a wavelength region unique to each direction from multiple directions, a camera for photographing the object, and the camera. A plurality of sensor sets having sensitivity only in the same wavelength range as the unique wavelength range of each of the plurality of spectral pattern projecting means; and a plurality of distance image calculating means for obtaining a distance image using image data obtained from each of the sensor sets. A plurality of distance image memories for storing distance image data obtained from each of the distance image calculation means; a distance image data integration means for integrating the contents of each of the plurality of distance image memories into one screen; A distance image memory for storing data output from the means.

〔Example〕

Next, the present invention will be described with reference to the drawings. First
FIG. 1 is a block diagram showing an embodiment of the present invention. In the figure, spectral pattern projection devices 32 and 33 simultaneously project spectral patterns of wavelengths λ _{1 to} λ ₂ and λ _{3 to} λ ₄ onto object 2, respectively. Here, λ ₂ <λ ₃ . The object 2 is photographed by the camera 31 and forms an image on the sensors 24, 25, 26, 27 via the light splitting elements 28, 29, 30. Image data obtained from the sensors 24 and 25 and the sensors 26 and 27 are input to distance image calculation units 34 and 35, respectively, to calculate distance images, and store the results in the distance image memories 36 and 37, respectively. The data of the distance image memories 36 and 37 are integrated into one screen by the integration processing unit 38, and the result is stored in the distance image memory 39.

Next, the operation of the distance image acquisition device shown in FIG.
This will be described with reference to FIGS. The wavelength ranges of the spectral patterns projected from the spectral pattern projectors 32 and 33 are λ _{1 to} λ ₂ and λ _{3 to} λ ₄ , respectively (λ ₂ <λ
₃ ). The object 2 is simultaneously illuminated using the spectral pattern projection devices 32 and 33, and the actual scene is photographed by the camera 31.
Sensors 24, 25, 26, and 27 built in camera 31 have spectral sensitivities σ ₃ , σ ₄ , σ ₅ , and σ ₆ as shown in FIGS. 2, ₃ , ₄ , and ₅ , respectively. Have. As is clear from the figure, the sensors 24 and 25 have sensitivity only to the wavelength range of the spectral pattern projection device 32, and the sensors 26 and 27 have sensitivity only to the wavelength range of the spectral pattern projection device 33. ing. Therefore, the sensors 24 and 25 detect only the spectrum pattern projected from the spectrum pattern projection device 32, and the sensors 26 and 27 detect only the spectrum pattern projected from the spectrum pattern projection device 33. No spectral pattern from the device is detected. Therefore, the outputs of sensors 24 and 25 and the outputs of sensors 26 and 27 are
Independently of each other, processing is performed in the distance image calculation units 34 and 35 using a known fact (JP-A-62-1953790), and each distance image can be obtained as follows.

First, prior to actual measurement, the diffraction position of the spectrum pattern is obtained. In FIG. 6, image data acquired by sensors having different spectral sensitivities σ ₃ and σ ₄ of the camera 31 are A / D
The ratio R is calculated for each pixel of the two image memories 9 and 10 via the converters 7 and 8 according to the equation (1), and the ratio image is stored in the ratio image memory 12.

When a white plate is placed on the plane 19 and the first calibration image is acquired, the function operation unit 14
A function X _a1 (R) of the ratio R on the plane 19 is obtained from the ratio image stored in 12 and held. The function X _a1 (R) has a value in the range of observed R _a11 ≦ R _a12 . Similarly, when a white plate is placed on the plane 20 and the second calibration image is obtained, the function calculating unit 15 calculates the ratio R of the plane 20 from the ratio image stored in the ratio image memory 12 via the switch 13. Function X
_b1 (R) is obtained and held. The function X _b1 (R) is given by R _a11 ≦
It has a value in the range of R ≦ _Ra12 .

The diffraction position calibration unit 16 calculates the functions X _a1 (R) and X _b1 (R)
For R in the overlapping part of the domain of both Rs,
According to equation (2), using an appropriate constant c, (X _a , Z _a ),
A pair of (X _b , Z _b ) is obtained, and a diffraction position (X ₀₁ , Z ₀₁ ) is obtained from Expressions (3) and (4). Calibrated diffraction position (X
₀₁ , Z ₀₁ ) are held in the diffraction position calibration unit 16.

Next, the diffraction position (X ₀₂ ,
Z ₀₂ ). Also in this case, the above-mentioned diffraction position (X ₀₁ , Z ₀₁ )
In the description of the method for determining _{the, σ 3, σ 4, X} a1 (R), X b1
(R), R _a11 , R _a12 , R _b11 , R _b12 , X ₀₁ , and Z ₀₁ are represented by σ ₅ ,
σ ₆ , X _a2 (R), X _b2 (R), R _a21 , R _a22 , R _b21 , R _b22 , X ₀₂ , Z
Just read ₀₂ and get exactly the same.

After obtaining the respective diffraction positions in this manner, the actual distance image measurement of the object 2 is performed.

When measuring the distance to the actual object 2, image data of a real scene acquired by sensors having different spectral sensitivities σ ₃ and σ ₄ of the camera 31 is transferred from the A / D converters 7 and 8 to the image memories 9 and 10. The obtained ratio image is obtained in the ratio image memory 12. Distance calculator
17 reads out the ratio R for each pixel from the ratio image memory 12 together with the X coordinate value via the switch 13, and stores the diffraction position (X ₀₁ , Z ₀₁ ) held in the diffraction position calibration unit 16 and the function calculation unit 14 , 15
Using the functions X _a1 (R) and X _b1 (R) held in Equation (5) or (6), the distance Z is obtained and written to the distance image memory 36.

As a result of the above-described operation in one row, the distance Z from the camera 31 to the object 2 is obtained for each pixel of the distance image memory 36.

Next, different spectral sensitivity sigma ₅ of the camera 31, image data obtained by the sensor of the sigma ₆ is above sigma _3, the description for the image data acquired by the sensor _{σ 4, σ 3, σ 4} ,
X ₀₁ , Z ₀₁ , X _a1 (R), X _b1 (R), and the distance image memory 36 are stored in σ ₅ , σ ₆ , X ₀₂ , Z ₀₂ , X _a2 (R), X _b2 (R), and the distance image, respectively. The processing is performed in parallel in exactly the same way just by reading the memory 37.

By the way, the distance images obtained in the distance image memories 36 and 37 measure the distance to the area A21 and the area B22, respectively. In the integration processing unit 38, a pixel value, that is, a measured distance is read for each pixel of the distance image memories 36 and 37, and the presence or absence of a measurement value is determined by the measurement value presence / absence determination unit 40. Transfers the value to the distance image memory 39. At this time, the distance of the area C42 measured in duplicate is:
The average value calculation unit 41 calculates the average value and transfers the result.

As a result of the above series of operations, a distance image in which the unmeasurable region caused by the blind spot of the spectral pattern is reduced by only one measurement is obtained in the distance image memory 39.

In the above-described example, the case where the number of sets of sensor sets corresponding to the spectral pattern projection device is two has been described. However, the case where the number of sets is three or more can be easily implemented in the same manner, and the same effect can be obtained. It is.

[Effects of the Invention] As described above, the present invention, from different directions,
A plurality of spectral pattern projection devices simultaneously project the spectral patterns unique to each device onto an object,
Simultaneous imaging with multiple sensor sets that have sensitivity only in the wavelength range unique to each spectral pattern, and calculating the distance image using the image data reduces the occurrence of spectral pattern blind spots even in a single measurement As a result, there is an effect that an unmeasurable region can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.
3, 4, and 5 are examples of the spectral impression of the sensor shown in FIG. 1, FIG. 6 is an example of a block diagram of the distance image calculation unit shown in FIG. 1, and FIG. FIG. 8 is an explanatory diagram of a conventional technique, and FIG. 9 is an explanatory diagram of a problem to be solved by the invention. 1,32,33 …… Spectral pattern projector, 2 …… Object,
3,31 …… Camera, 5,6,24,25,26,27 …… Sensor, 7,8 ……
A / D converter, 9,10… Image memory, 11… Ratio image operation unit, 12… Ratio image memory, 13… Switch, 14,15…
Function calculation unit, 16: diffraction position calibration unit, 17: distance calculation unit, 18, 36, 37, 39: distance image memory, 19, 20: plane,
21 ... Area A, 22 ... Area B, 23 ... Camera field of view, 4,2
8, 29, 30 ... optical branching element, 34, 35 ... distance image calculation unit, 38
... Integration processing unit, 40... Measurement value presence / absence determination unit, 41... Average value calculation unit, 42.

Claims (1)

(57) [Claims] 1. A distance image acquisition device for projecting a spectral pattern that has been separated onto an object, capturing the image with a camera having sensors having different spectral sensitivities, and obtaining a distance to the object for each pixel. The spectral pattern of the wavelength range specific to each direction is
A plurality of spectral pattern projecting means for simultaneously projecting the object, a camera having a plurality of sensor sets having sensitivity only to a unique wavelength range of each of the plurality of spectral pattern projecting means, A plurality of processing means corresponding to each sensor set for obtaining a distance to an object based on outputs of the plurality of sensor sets, and integrated processing means for integrating a plurality of distance images obtained from the plurality of processing means into one distance image A distance image acquisition device, characterized in that: