JP3339221B2 - Surface direction detector - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for measuring the direction of a mirror surface by utilizing the property of regular reflection of light on the mirror surface.

[0002]

2. Description of the Related Art A method of detecting the surface direction of a mirror surface by using the property of specular reflection of light on the mirror surface is disclosed in Japanese Patent Application Laid-Open No. H10-157,197. It is determined whether or not the light corresponds to the light, and in the regular reflection, the fact that the incident angle and the reflection angle are equal is used to determine the surface direction of the reflection surface from the position of the light source.

In this prior art, in order to specify a light source of reflected light, a plurality of light sources are sequentially turned on to detect the presence or absence of reflected light, and the position of the light source emitting the detected reflected light is determined. What you want (eg ACSanderson, LEWeiss
and SKNayar: "Structuredhighlight inspection of
specular surfaces ", IEEE Trans. Pattern Anal. & Mac
h. Intell., PAMI-10, 1, pp.44-55 (1988)). But,
In this method, since the light sources are sequentially turned on to detect the reflected light, there is a problem that the measurement takes time. In particular, there is a problem that the measurement time becomes longer in proportion to the type of the surface direction to be detected.

As an improvement of this method, light sources having different wavelengths (colors) are used at different positions, all the light sources are turned on at the same time, light reflected from an object is received, and the wavelength of the reflected light is detected. There is a method of specifying a position of a light source by a wavelength. This method can reduce the measurement time because all light sources are turned on at the same time (for example, Seiji Hata and Masako Nishiyama: "Extraction of mirror surface shape by stereo vision under color stripe illumination", the 8th session) Proceedings of the Image Sensing Technology Symposium in Industry, pp.103-108 (1992-
6)). However, in this method, it is necessary to detect the reflected light by dividing it into wavelengths, so that there has been a problem that the device configuration becomes complicated.

[0005] Further, a transmission type surface light source is produced using a projection optical system, this is placed sufficiently close to an object, and information on the normal direction and depth of the object surface is obtained under the constraint conditions for the object. There is a method (Etsuji Nishino, Yoshiaki Shirai, "Determination of metal surface shape by photometric stereo method using projection optical system",
Computer Vision 32-2, (1984, 7)). This method uses a filter whose transmittance changes in the form of a wedge, projects the light on a diffuser, and makes the distribution of brightness on the diffuser a wedge, and uses this diffuser as a surface light source to reflect light from the diffuser. The object is irradiated with light. Then, the reflected light from the object is received by the television camera, and the light source position on the surface light source corresponding to each pixel on the image is once determined based on the density information (reflected light intensity) on the light receiving surface of the television camera. Seeking to.

However, in order to avoid the influence of the reflectivity of the object surface, a surface light source of almost uniform brightness, which is produced by projecting onto a diffusion plate with a projection optical system without a filter, is used. Light intensity is standardized. In this method, the influence of the reflectance of the object surface is not affected by the normalization, but since the reflected light intensity is basically used as it is, it is easily affected by the noise of the camera. Therefore, there is a problem that the detection accuracy in the plane direction is not good because the accuracy of determining the corresponding point is not good.

That is, in the conventional method using the projection optical system,
Since the normalized light intensity is standardized by the reflected light intensity of the surface light source having substantially uniform brightness, the ratio of the light emission intensity is distributed from 1 / A to 1 (A> 1) at each position. For this reason, since the dynamic range of the light emission intensity ratio is narrow, there is a problem that it is easily affected by the noise of the camera and the detection accuracy in the plane direction is not good.

[0008]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to reduce the measuring time and simplify the apparatus configuration when measuring the surface direction of an object. To improve the detection accuracy.

[0009]

According to an aspect of the present invention, there is provided an apparatus for detecting a surface direction of an object by using reflected light, wherein a distribution of a light emission intensity of a light emitting unit according to a position in the light emitting unit is determined. , A> 1, as a normalized value
The pattern can be changed from approximately 1 / A to 1 and from 1 to approximately 1 / A, and the distribution of the ratio of the two emission intensities for each position with respect to the position monotonically increases from approximately 1 / A to A. Illuminating means for irradiating the object in two patterns, and the intensity of the reflected light from the object of the incident light irradiating the object from a position of the light emitting unit of the illuminating means,
For each of the two patterns, the radiation position of the incident light beam corresponding to the reflected light beam on the light emitting portion is determined from the measuring means for measuring and the intensity ratio of the two reflected light beams to the two patterns measured by the measuring means. And calculating means for obtaining the surface direction of the object based on the radiation position.

According to another aspect of the present invention, a plurality of lighting means are provided in a row,
Each illumination unit emits incident light used for detecting each of the divided angular ranges of the measurable range in the plane direction of the object. According to another aspect of the invention, a large number of illumination means are provided along the inner surface of a hemisphere or a cone with the optical axis of the measurement means as a central axis, and each illumination means has a divided angle of a measurable range in the plane direction of the object. It emits incident light used for detecting a range.

[0011]

The lighting means can emit light with the light emission intensity of each light emitting portion as a first pattern and a second pattern. The emission intensities of the first and second patterns are adjusted so that the emission intensity ratio between the first and second patterns monotonically increases in the range of 1 / A to A. However, it is not necessary to light emission intensity ratio in relation to the light emission position increases monotonically, it is meant that not take the same value as the value.

[0012] By such an illumination means, the first object is placed on the object.
The light is illuminated by the pattern, and the intensity of the reflected light from the object is measured by the measuring means. This measuring means is adjusted so that it can receive only the reflected light corresponding to the light emitted from one position on the light emitting surface at the same reflection position of the object. The measured value of the reflected light intensity at this time is defined as S1. Similarly, the object is irradiated with light according to the second pattern, and the intensity of the reflected light is measured. The measured value of the reflected light intensity at this time is defined as S2. Then, the ratio S1 / S of the above reflected light intensity is obtained.
2 is calculated, and the position on the light emitting surface where the ratio of the light emitting intensity is closest to S1 / S2 is calculated between the first pattern and the second pattern in the light emitting intensity of the light emitting surface. Then, the direction of the reflecting surface of the object is calculated from the relationship between the optical axis of the light incident on the object at the position on the light emitting surface and the optical axis of the reflected light detected by the measuring means.

As described above, the object can be measured in the plane direction by irradiating the object with the light emission intensity pattern of the illumination changed in two ways and measuring the ratio of the intensity of the reflected light from the object. Therefore, there is an effect that the measurement time can be reduced. Further, since the wavelengths of the light sources may be the same, there is an effect that the apparatus is simpler than when the wavelength is changed. Since the ratio of the reflected light intensity is used, the surface direction can be detected without being affected by the reflectance of the surface of the object or by the absolute change in the light emission intensity of the light source of the illumination means. Further, in the first pattern and the second pattern, the distribution of the light emission intensity ratio on the light emitting portion is substantially 1 unit.
Since the function is monotonically increasing between / A and A, the dynamic range is widened and the corresponding point can be determined with high accuracy as compared with the conventional method in which the reflected light intensity of the surface light source having almost uniform brightness is standardized. Thus, the detection accuracy in the plane direction is improved.

By providing a plurality of lighting means in a row,
The detectable angle range can be divided, and the resolution of detection in the plane direction can be improved. Further, by providing a large number of illumination means along the inner surface of a hemisphere or a cone with the optical axis of the measurement means as the central axis, it is possible to detect the azimuth of the surface facing any direction.

[0015]

EXAMPLES first embodiment of the first embodiment the present invention shown in FIG. This apparatus measures the plane direction of a plane on which a normal vector exists on the paper surface. In FIG. 1, 1 is an object to be measured, 2 is an LED lighting device in which LEDs (light emitting diodes) are arranged in a straight line, 3
Is a television camera for detecting the distribution of the intensity of reflected light from the object 1 illuminated by the LED illumination device 2 as an image,
4 is a camera lens of the television camera 3, 5 is an image input device for inputting an image taken by the television camera 3 to the computer 7,
Reference numeral 6 denotes an LED driving device that drives the LED lighting device 2, 7 a computer that controls the LED driving device 6 and calculates a surface direction distribution of an object from image data input from the image input device 5, 8 denotes a television camera 3 And a pedestal for fixing the LED lighting device 2. Note that the LED lighting devices 2 may be arranged on an arc as shown in FIG. 7, for example, in order to equalize the size of each divided angle range. Since the TV camera 3 and the LED lighting device 2 are fixed to the same pedestal, by using a camera lens having a small depth of focus,
By adjusting the position of the pedestal so that the television camera is in focus, it is possible to position the object 1 with the television camera 3 and the LED lighting device 2.

FIG. 2 shows the configuration of the LED lighting device 2.
In FIG. 2, LEDs 1 to 12 are LEDs, and the cathode side of each LED is connected to a terminal G. R1,1 to R1,12 are resistors. One terminal of each resistor is connected to the anode of LED1 to LED12, and the other terminal is connected to terminal V1. R2,1 to R2,12 are resistors, one terminal of each is LED1 to LED1
2 and the other terminal is connected to terminal V2.

The resistances of the resistors R1,1 to R1,12 ideally change in an equal ratio. Also, the positive resistance
The resistance values of R2,1 to R2,12
2 is equal to the resistance value of R1,1. Since it is difficult to obtain a resistor whose resistance value changes strictly and equidistantly, for example, an E-12 series or E
-24 series resistance is used. FIG. 3 shows that the resistors R1,1 to R
1,12 and R2,1 to 2,12 are shown. The light emission intensity of the LEDs 1 to 12 is proportional to the drive current within the range of the drive current of the LED determined by these resistance values. Therefore, the emission intensity of each LED when the resistors R1,1 to R1,12 are used increases geometrically, and the resistors R2,1
Therefore, the emission intensity of each LED when R2,12 is used decreases geometrically.

First, the principle of obtaining the angle of the surface will be described.
FIG. 4 shows a case where a constant voltage V is applied between the terminal V1 and the terminal G of the LED lighting device 2 (first pattern) and a case where the terminal V2
7 shows the light emission intensity distributions L1 (i) and L2 (i) in one axis direction of the LED when a constant voltage V is applied between the LED and the terminal G (second pattern). In addition, two emission intensities L1 (i)
FIG. 5 shows a ratio L1 (i) / L2 (i) between the ratio L1 (i) and L2 (i). Here, i is the identification number of the LED, i = 1, 2,.
, 12 are LED1, LED2, ..., LE respectively
Corresponds to D12.

When the light from the i-th LED is specularly reflected on the surface of the object and is incident on the television camera, the light is incident on the television camera on the same optical path in both the first pattern and the second pattern. I do. Therefore, when the reflectance on the surface of the object is r, the intensity of light incident on the television camera is r · L1 (i) for the first pattern L1 (i),
R · L2 (i) for the pattern L2 (i).
Therefore, the ratio of the intensity of light incident on the television camera is given by:
1 (i)｝ / {r · L2 (i)}, that is, L1 (i) /
L2 (i), which is equal to the above-described ratio L1 (i) / L2 (i) of the emission intensity of the i-th LED.

Thus, if the intensity ratio of the specularly reflected light to the television camera 3 is known, the light emission intensity ratio of the LED emitting the incident light corresponding to the reflected light can be known. As shown in FIG. 5, the LED identification number i
Is uniquely determined, so that it is possible to know from which LED the specularly reflected light from the surface of the target object 1 is emitted. When the light emission intensity at the same drive current differs for each LED due to the variation between the LED elements and the dirt on the LED light emission surface, the distribution characteristic of the light emission intensity is as shown in FIG. However, also in this case, the emission intensity ratio of each LED is as shown in FIG. 9 and is equal to that of FIG.

FIG. 6 shows the principle that the angle of the surface is determined from the position of the LED of the light source which becomes the regular reflection light from the surface of the object 1. When the direction of the incident light B _i to LED radiation relative to the optical axis z of the television camera 3 and θa degree, the distance between the object 1 and the LED lighting device 2 of the measurement region of the object 1 size and a camera lens When the aperture is sufficiently larger than the opening, the angle θ parallel to the paper surface in the plane direction (vector perpendicular to the plane) is θa / 2 degrees as shown in the figure.

According to the present method, the ratio of the light emission intensity of the LED, which is the information for identifying the LED, is L1 (i) / L2 (i, as compared with the conventional method of normalizing with uniform brightness. )
The difference between the LEDs increases. That is, the geometric coefficient of the light emission intensity of the LED, that is, the geometric coefficient C of the resistance value (C
> 1), the ratio of light emission intensity L1 (i) / L2
The minimum change in (i) is C times in the conventional method, but is 2 times C in the present method. Therefore, according to the present method, the influence of the noise of the TV camera is reduced, so that the corresponding point can be determined with higher accuracy than in the conventional method in which the reflected light intensity of the substantially uniform surface light source is standardized, and the surface direction can be determined. The detection accuracy is improved.

Next, the operation of the apparatus shown in FIG. 1 will be described. First, the computer 7 controls the LED driving device 6 to apply a constant voltage V between the terminals V1 and G of the LED lighting device 2, thereby causing the LED lighting device 2 to emit light in the first pattern. The image of the object 1 at this time is captured by the television camera 3, and the image data D 1 (x, y) is taken into the computer 7 by the image input device 5.

Next, the computer 7 controls the LED driving device 6 to apply a constant voltage V between the terminals V2 and G of the LED lighting 2, thereby causing the LED lighting device 2 to emit light in the second pattern. At this time, the image of the object 1 is captured by the television camera 3, and the image data D2 (x, y) is similarly loaded into the computer.

The image data D1 (x, y) and D2
In (x, y), the pixel value is proportional to the logarithm of the intensity of light incident on the television camera 3. This can be easily realized by a process of converting the output value of the television camera 3 using a look-up table calibrated in advance. The pixel value is converted so as to be proportional to the logarithm of the incident light intensity so that the ratio of the light intensity can be obtained by subtraction.

Thus, the data acquisition has been completed. Next, the computer 7 calculates the surface direction distribution. As described above, since the pixel value of the image data D1 (x, y) and D2 (x, y) is proportional to the logarithm of the intensity of light incident on the television camera 3, D1 (x, y) -D2 (x, y) y) is k · l
og (L1 (i) / L2 (i)). Here, k is a proportionality constant. Therefore, D1 (x, y) -D2 (x,
The angle θ (x, y) is obtained for each pixel (x, y) from y). However, what is required here is the object 1,
Pixels at angles within an angle measurement range determined by the positional relationship between the television camera 3 and the LED lighting device 2. If any of the pixel values of the image data D1 (x, y) and D2 (x, y) is equal to or less than a predetermined threshold value, there is a strong possibility that reflected light other than a mirror surface or noise of the television camera 3 will cause noise. , Ignored as invalid data.

The angle θ is D1 (x, y) −D2 (x,
A look-up table from the value of y) to the angle can be created in advance, and the value can be obtained by subtracting this table. This lookup table is based on the object 1, L
A method of creating by a geometric calculation from the position of each LED of the ED lighting device 2 and the television camera 3, and a method of creating based on image data taken by actually placing various angles on the position of the object 1. There is a way to do it. Through the above processing, data of the angle θ (x, y) is obtained.

In this embodiment, the LEDs are arranged on a straight line because the surface direction of the object to be measured has only a component parallel to one plane including the optical axis z of the television camera. For example, to measure a surface facing various directions, the first,
Instead of arranging a plurality of LEDs emitting light in the second pattern on a straight line, for example, as shown in FIG.
(LED1, LED2,..., LEDn) enables the detection of the direction of the surface facing any direction.

That is, the vector from the object 1 to the LEDi is represented by F
^* i, when the optical axis vector of the TV camera is represented by Z ^* , LEDi
Gives the normal vector N ^* as (F ^* i / | F ^* i | + Z ^* / |
It is possible to detect the surface Ti that is Z ^* |) / 2. Also, LEs arranged on a straight line as shown in FIG. 1 or FIG.
By scanning the D illumination device relative to the object, it is also possible to detect surfaces facing various directions.

Further, it is apparent that by repeating the measurement according to the present embodiment at a high speed, it is possible to measure the temporal change of the distribution of the object in the surface direction. In this embodiment, the LED is used as the light source, but another method such as a projection optical system may be used.

In this embodiment, the number of LEDs constituting the LED illumination is set to twelve. The number of LEDs depends on the required measurement accuracy, and on the other hand, the dynamic range of the television camera 3 and the S / N ratio. That is, when the number of LEDs is 12, the number of surface directions that can be measured is 12 at the maximum. If the type of the reflected light intensity ratio that can be sensed by the television camera 3 is at most 12, L
The maximum number of EDs is 12. If the number of LEDs is further increased in terms of required measurement accuracy, for example, the second
Embodiments can be used. In addition, by using the television camera 3 having a wider dynamic range, the number of types of reflected light intensity ratios that can be sensed can be increased, so that the number of LEDs can be increased, and thus measurement accuracy can be increased.

[0032] A second embodiment of the second embodiment the present invention in FIG. 11. The present embodiment is a solution in the case where the number of LEDs cannot be increased due to the dynamic range and S / N of the television camera 3. Here, in order to increase the angle measurement accuracy, L
A case where the number of EDs is set to 24, which is twice that of the first embodiment, is shown.
In the device of the present embodiment shown in FIG. 11, the LED lighting device 2 includes two systems of LED lighting devices 2A and 2B. The LED lighting devices 2A and 2B have the same configuration as the LED lighting of the first embodiment. LED lighting device 2A
The configuration is the same as that of the first embodiment except for 2B and 2B. Note that L
The ED lighting devices 2A and 2B may be arranged on the same arc as in FIG.

Next, the operation of the apparatus shown in FIG. 11 will be described. In the present embodiment, the LED lighting devices 2A and 2B are separately turned on and an image is taken by the television camera 3. In the present embodiment, the detection angle range is divided into two, and one of the ranges is set to the LED lighting device 2.
A, and the other range is detected by the LED lighting device 2
This is detected by the illumination of B. Since the LED lighting devices 2A and 2B have 12 types of emission intensity ratios,
After all, this is equivalent to detecting the entire detection angle range with 24 light beams.

First, the LED lighting device 2A emits light in the first pattern, an image of the object 1 is picked up by the television camera 3, and the image data D1A (x, y) is taken into the computer 7. Next, the LED illumination device 2A emits light in the second pattern, an image of the object 1 is captured by the television camera 3, and the image data D2A (x, y) is taken into the computer 7. Next, the LED illumination device 2B is caused to emit light in the first pattern, and an image of the object 1 is captured by the television camera 3,
The image data D1B (x, y) is taken into the computer 7. Next, the LED lighting device 2B emits light in the second pattern, an image of the object 1 is captured by the television camera 3, and the image data D2B (x, y) is taken into the computer 7. The image data D1A (x, y), D2A (x, y), D1B (x, y)
y), D2B (x, y) are those whose pixel values are
Is proportional to the logarithm of the intensity of the incident light to the light source.

Next, the computer 7 calculates the distribution in the plane direction. The angle θ (x, y) is obtained for each pixel (x, y) from D1A (x, y) −D2A (x, y). However, what is required here is the object 1, the TV camera 3
And a pixel having an angle within an angle measurement range determined by the positional relationship between the LED and the LED lighting device 2A. Next, image data D1B (x, y)
And D2B (x, y) perform the same processing to determine the angle. However, what is required here is the image data D
Of the pixels whose angles were not determined from 1A (x, y) and D2A (x, y), the object 1, the TV camera 3 and the LED
Pixels at angles within the angle measurement range determined by the positional relationship of the lighting device 2B. By the above processing, data of the angle θ (x, y) of the pixel having the angle in both measurement ranges is obtained.

In this embodiment, the LED lighting device is LE
Although this is a case where the D lighting device is divided into two systems of the D lighting devices 2A and 2B, it is apparent that a configuration in which the LED lighting device is divided into three or more systems is also possible if necessary.

In each of the above embodiments, the LED elements having the light emission intensity are arranged in the order in which the light emission intensity monotonously increases or decreases. However, the arrangement order of the LED elements is not limited to this. For example, in the case of the lighting device of FIG.
The geometrical arrangement order of each LED having a known emission intensity is LED1, LED5, LED1 in order from the left.
2,..., LED3. In short, the effects of the present invention can be achieved if the relationship between the light emission intensity and the light emission intensity ratio and the light emission position is known in advance.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a surface direction detection device according to a first embodiment of the present invention.

FIG. 2 is a circuit diagram illustrating a configuration of a lighting device.

FIG. 3 is an explanatory diagram showing resistance values connected to each LED of the lighting device.

FIG. 4 is a characteristic diagram showing a distribution of light emission intensity in a first axis direction in a first pattern and a second pattern of the lighting device.

FIG. 5 is a characteristic diagram showing a distribution of a ratio of light emission intensity in a first pattern and a second pattern of the lighting device in one axis direction.

FIG. 6 is an explanatory diagram showing a principle of detection in a plane direction.

FIG. 7 is a configuration diagram showing a modification of the illumination device of the first embodiment.

FIG. 8 is a characteristic diagram showing the distribution of the luminous intensity in the first pattern and the second pattern in the direction of one axis when the luminous intensity of the LED of the lighting device is uneven.

FIG. 9 is a characteristic diagram showing the distribution of the ratio of the luminous intensity in the first pattern and the second pattern in the case of non-uniform luminous intensity of the LEDs of the lighting device in one axial direction.

FIG. 10 is a configuration diagram showing a method of arranging LEDs of another lighting device of the first embodiment.

FIG. 11 is a configuration diagram showing a surface direction detection device according to a second embodiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Target object 2 ... LED lighting device 3 ... TV camera 4 ... Lens 5 ... Image input device 6 ... LED drive device 7 ... Computer

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-4-212047 (JP, A) JP-A-6-137826 (JP, A) JP-A-3-142303 (JP, A) (58) Investigation Field (Int.Cl. ⁷ , DB name) G01B 11/00 G01C 3/06

Claims (3)

(57) [Claims] 1. The distribution of the luminous intensity of the luminous part according to the position in the luminous part , where A> 1, is approximately a normalized value.
It is possible to change from 1 / A to 1 and from 1 to almost 1 / A in two patterns, and the distribution of the ratio of the two emission intensities for each of the positions with respect to the position is monotonically in the range of approximately 1 / A to A. As an increasing function, an illuminating means for irradiating the object with the two patterns, and the intensity of the reflected light from the object of the incident light radiated to the object from a position of the light emitting unit of the illuminating means Measuring means for measuring the two patterns, respectively; and measuring the light emission of the incident light corresponding to the reflected light from the intensity ratio of the two reflected lights to the two patterns measured by the measuring means. Calculating means for determining the radiation position on the part and determining the surface direction of the object based on the radiation position. 2. The lighting device according to claim 1, wherein a plurality of the illuminating units are provided in a row, and each of the illuminating units emits incident light used for detecting each of the divided angular ranges of the measurable range in the surface direction of the object. The plane direction detecting device according to claim 1, wherein: 3. A plurality of illuminating means are provided along an inner surface of a hemisphere or a cone centered on an optical axis of the measuring means, and each illuminating means measures a surface direction of the object. The surface direction detecting device according to claim 1 or 2, wherein the device emits incident light used for detecting each of the divided angle ranges of the possible range.