JPH08313226A - Method of inspecting transparent object - Google Patents

Abstract

PURPOSE: To provide an inspection method in which the shape of a transparent object can be grasped even when the shape is irregular. CONSTITUTION: Pattern light 1 from a pattern-light projector 2 is projected onto a transparent object 4 on an opaque object 3. An imaged pattern which is monitored by a TV camera 5 is subjected to the influence of the transparent object 4 so as to be deformed. On the other hand, a pattern which may be obtained at a time when the pattern light is projected onto an initial model, in a three-dimensional shape, which has been prepared in advance is computed. Then, a pattern which is close to the actually imaged pattern is selected, a new model in which a model corresponding to them has been composed is created, and the shape of the transparent object 4 is estimated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for inspecting a transparent object, and more particularly to a method for non-contact acquisition using an optical method.

[0002]

2. Description of the Related Art As a method for inspecting an accurate shape of a transparent object (including an object that is almost transparent here) in a non-contact manner, it has been attempted to use a conventionally developed method such as a light-section method or stereo vision Then, most of the projected light rays are transmitted through the surface of the transparent object and very few light rays are reflected by the surface of the transparent object, so that they cannot be applied.

[0003]

SUMMARY OF THE INVENTION Therefore, the present invention provides a method capable of accurately acquiring a shape of a transparent object which could not be inspected by a conventional inspection method, in a non-contact manner, and in particular, a transparent object. The purpose is to be able to grasp the shape even if the shape is irregular.

[0004]

In order to solve the above-mentioned problems, a method for inspecting a transparent object according to the present invention comprises projecting pattern light onto a transparent object to be inspected arranged on an opaque object, and inspecting the object to be inspected. After being projected on the opaque object through the first step, a first step of obtaining an imaging pattern monitored through the object to be inspected again, and a portion showing a region where the object to be inspected in the monitored imaging pattern is a bottom surface A second step of creating a large number of three-dimensional initial models, a third step of creating a sample pattern that will be monitored in each obtained model when the pattern light is irradiated, Fourth step of comparing each sample pattern with the imaging pattern and selecting some of models corresponding to sample patterns close to the imaging pattern A fifth step of combining the selected model, a model for the combined three-dimensional shape performed again repeated the third, fourth step,
The sixth step of acquiring the shape of the object to be inspected.

[0005]

When the shape of the transparent object is irregular and unpredictable, the shape cannot be analytically obtained from the deformation amount of the imaging pattern. Therefore, in the above method, an initial model that is a large number of samples is generated in advance, and an imaging pattern that would be obtained when the pattern light is projected in the initial model is calculated and obtained as a sample pattern. Then, some sample patterns close to the actually obtained imaging pattern are selected, and the models corresponding to them are synthesized to create a new model.

A sample pattern is calculated for this new model, a sample pattern closer to the actual image pickup pattern is selected, and a new model is synthesized.
In this way, the shape of the test object is predicted.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The method for inspecting transparent objects according to the present invention will be specifically described below with reference to the drawings. FIG. 1 shows a first embodiment of the present invention. In the figure, a pattern light projector 2 outputs a pattern light 1 (a parallel slit light in this case) to a transparent object 4 arranged on an opaque object 3. To project. The projected pattern light 1 is refracted on the surface of the transparent object 4 to form a deformed pattern on the opaque object 3. This deformation pattern is photographed by an image pickup device such as the TV camera 5, and the deformation state of the pattern is taken into the image processing device 6. The image pickup pattern monitored by the TV camera 5 is, as shown in FIG.
The image pickup pattern 8 in which the region in which is present is deformed is displayed.

The deformed pattern light on the opaque object 1 is T
When moving toward the V camera 3, it is refracted again by the surface of the transparent object 4, but since the surface of the opaque object 1 has some diffuse reflection component, the position of the TV camera 3 is generally not specified and observation is not possible. It is possible. When the transparent object 4 has an irregular shape, the transparent object 4
It is difficult to analyze the shape of the. Therefore, in the present embodiment, sample patterns obtained by projecting pattern light on various three-dimensional samples prepared in advance are used,
The shape of the transparent object 4 is estimated, and the method will be described below with reference to the flowchart shown in FIG.

In step S0, an image pickup pattern obtained by projecting the pattern light on the transparent object 4 which is the object to be inspected is input to the image processing device 6. In step S1,
The existence range of the transparent object 4 is determined from the imaging pattern 7.
This determines the existing position in the screen when the transparent object 4 is viewed from above, analyzes the image displayed on the TV monitor 7, and first extracts the linear pattern of the peripheral portion where there is no transparent object. Then, a portion having a large deviation is extracted as the outer circumference of the transparent object 4.

In step S2, an initial model is generated within the existence range of the transparent object 4. The initial model is shown in Figure 4.
As shown in (a), the existence range is defined as a bottom surface, and a point 9 of height H taken at a randomly determined position within the range is defined as a vertex and a cone formed by connecting the edges of the bottom surface. Further, as shown in FIG. 4B, this is converted into the height of each point in the transparent object region, and smoothing is applied to the whole to form a smooth surface. The position of the apex 9 is changed, and a large number of such initial models are generated.

Next, in step S3, an image pickup pattern which will be obtained when the pattern light is projected on these initial models is calculated. This calculation process consists of the following three processes (A), (B) and (C). (A) In this process, the position on the opaque object 14 where the slit light projected from the projector is projected is calculated. This calculation procedure will be described with reference to FIG. Pattern light projector 11 for model 10
It is considered that the three-dimensional arrangement is arranged on the computer so that it becomes similar to the configuration shown in FIG. Next, a large number of slit lights 12 projected from the pattern light projector 11
All are sampled at appropriate intervals, and the slit light 12 is made into a set of many light rays 13. This ray 13
Is refracted at the surface of the model 10 and is projected to a position 15 on the opaque object 14 or directly to a position 15 on the opaque object 14 without passing through the model 10. The projected position 15 is calculated for all the light rays 13 and used as projection position data.

(B) In this process, the position on the opaque object 14 where the line of sight corresponding to each pixel of the image pickup camera reaches is calculated. This calculation procedure will be described with reference to FIG. It is considered that the imaging camera 16 is arranged on the computer so that the three-dimensional arrangement of the imaging camera 16 is the same as that shown in FIG. The imaging camera 16 has a plurality of pixels 17 and has a line of sight 18 corresponding to each of the pixels 17.
There is. Each line of sight 18 is refracted at the surface of the model 10 to reach a certain position 19 on the opaque object 14 or directly to a certain position 19 on the opaque object 14 without passing through the model 10. By collating the position 19 with the projection position data 15 (described in (C)), it is possible to determine whether or not the line of sight 18 captures the light beam 12 projected from the pattern light projector 11.

(C) In this process, it is determined whether or not the position 15 where the light beam 12 is projected can be captured by the line of sight 18. This procedure will be described with reference to FIG. On the surface of the opaque object 14, the projection positions 15 of the light beam 13 are discretely distributed. A line of sight 1 on this opaque object 14
If 4A reaches the position 19A, it is assumed that the sight line 14A catches the light ray 12 because the projection position 15 exists around the position 19A. If another line of sight 14B reaches the position 19B, it is assumed that the line of sight 14A does not capture the light beam 12 because there is no projection position 15 around the position 19B.

The above operations (B) and (C) are performed for all the pixels 17, and the pixels 17 corresponding to the line of sight 18 that captures the light ray 12 are made bright, and the other pixels are made dark so that they are calculated from the model 10. It is possible to obtain the image pickup pattern. Next, in S4, the actual imaging pattern and S3
The imaging pattern of each model calculated in step 1 is evaluated for each line of the slit light, and a fraction of the upper number of the initial model with a small error is selected. FIG. 6 shows the evaluation method of the imaging pattern. For each line of slit light on the opaque object 14,
A shift amount 18 between the line 16 of the actual image pickup pattern and the image pickup pattern 17 of the model is calculated. This value is calculated for all the lines of the pattern, and the sum is taken to be the evaluation value of the model.

In S5, when the evaluation value of the model selected in S4 is smaller than the preset reference value, this model is made into the shape of a transparent object. However, in most cases, this criterion is not satisfied, so the model is deformed in S6. In S6, a number of new models are created by combining the plurality of models selected in S4. This preparation method is shown in the plan view of FIG. Two models 20 and 21 are selected from the plurality of models selected in S4, and then divided into four at the same position by randomly generated vertical and horizontal lines, and one of them is randomly replaced, and a new one is created. This is model 22. This is combined among multiple models to obtain many new models. At this time, a step is generated at the joint, so the model is smoothed by smoothing the height.

In some of the models combined as described above, the shape of the object surface is slightly changed and smoothed by smoothing to deform the model. With respect to this combined and deformed large number of models, S3,
Better approximation can be performed by performing S4 and selecting some models with good evaluation. This S6, S
If the processing of S3 and S4 is performed under the condition of S5, that is, if the evaluation value is smaller than the reference value, a model approximating the actual shape of the transparent object 4 can be obtained on the image processing apparatus. In addition,
In S5, if the number of times the steps S3 to S6 are repeated exceeds the specified number, the process can be aborted to shorten the calculation time.

[0017]

As described above, according to the present invention, even if the transparent object has an irregular shape, the shape can be grasped.

[Brief description of drawings]

FIG. 1 is a perspective view showing an embodiment of an inspection apparatus using the transparent object inspection method of the present invention.

FIG. 2 is a diagram showing an imaging pattern monitored in the same apparatus.

FIG. 3 is a flowchart showing an inspection process of the transparent object inspection method of the present invention.

FIG. 4 is a perspective view showing an initial model created by the same method.

FIG. 5 is an explanatory diagram of a process of improving the model in the same method.

FIG. 6 is an explanatory diagram of a process of evaluating a model in the same method.

FIG. 7 is a plan view illustrating a process of synthesizing models in the same method.

[Explanation of symbols]

 1,12 pattern light 2,11 pattern light projector 3,14 opaque object 4 transparent object 5 TV camera 6 image processing device 7 TV monitor 8 imaging pattern 10 model 17 pixels

Claims (1)

[Claims] 1. An imaging pattern in which pattern light is projected on a transparent test object arranged on an opaque object, projected on the opaque object through the test object, and then monitored again through the test object. And a second step of creating a large number of three-dimensional initial models whose bottom surface is a portion indicating the existence area of the object to be inspected in the monitored imaging pattern, and in each obtained model, A third step of calculating a sample pattern that will be monitored when the pattern light is irradiated is compared with each of the sample patterns and the imaging pattern, and corresponds to a sample pattern close to the imaging pattern. A fourth step of selecting some of the models; a fifth step of combining the selected models; For the three-dimensional shape model of again the third,
A method of inspecting a transparent object, comprising: a sixth step of repeatedly performing the fourth step to acquire the shape of the object to be inspected.