JP2712884B2 - Work identification method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for identifying a workpiece, and more particularly, to a method for identifying whether or not a workpiece is a target workpiece based on an image of the workpiece obtained by an image detection unit. And its device.

[0002]

2. Description of the Related Art Conventionally, it has been proposed to use a feature value to identify whether or not a work is a desired work. Here, as feature values that may be generally adopted,
Number of holes, number of corners, area, total area, maximum hole area, minimum hole area,
Thirteen types of characteristic values are known, such as a perimeter, a maximum radius, a minimum radius, a major axis length of an inertia ellipse, a minor axis length of an inertia ellipse, a circumscribed rectangle width, and a circumscribed rectangle length. Therefore, by obtaining variances for these 13 types of feature values, and combining some feature values to determine whether or not the work is a desired work, it is possible to achieve work identification with extremely high accuracy.

[0003]

However, obtaining a variance for each of the thirteen types of feature values, and combining several feature values to determine whether or not the work is a desired work causes a significant increase in processing load. As a result, there is an inconvenience that the time required for workpiece identification is significantly increased. Therefore, due to the tact time, it cannot be put to practical use in the field of component assembling where improvement in processing speed is strongly required.

In order to solve such inconveniences, the present inventors weighted the 13 types of feature values equally, selected some arbitrary feature values, and selected the selected feature values. Considering the identification of the work based on the value, actual identification of the work was performed. However, the accuracy of the identification of the work was limited, and depending on the type and state of the work, there were cases where the identification could not be performed sufficiently. That is, there is no guarantee that a feature value suitable for the geometric feature of the work to be identified is necessarily selected when selecting any arbitrary feature value, and a feature value that is not appropriate is selected. As a result, the work identification accuracy is reduced as described above. If the work identification accuracy is reduced, for example, when assembling parts, it is more likely that an incorrect work will be assembled, so that visual judgment by the operator must be intervened. However, it cannot be practically used for work identification in an automated line.

[0005]

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and provides a work identification method and apparatus capable of reducing the number of characteristic values for work identification and significantly improving the accuracy of work identification. It is intended to be.

[0006]

According to a first aspect of the present invention, there is provided a work identification method for extracting a feature value of a plurality of teaching works, and extracting an area value and a feature value of the extracted feature values. Select a value that defines the surrounding contour, and set the area value and the value that defines the surrounding contour when the accuracy of identifying the workpiece by the area value corresponding to the type of the teaching work and the value that defines the surrounding contour is high. Select the feature value for work identification, select the feature value with the highest identification accuracy when the work identification accuracy is low, as the work identification feature value, and set the upper and lower limits based on the average and variance of the selected feature values. This is a method of obtaining a threshold and identifying a target workpiece based on the selected work identification feature value and upper and lower thresholds.

According to a second aspect of the present invention, there is provided a work identification apparatus for extracting a characteristic value of a plurality of teaching works, and selecting a value defining an area value and a surrounding contour from the extracted characteristic values. First selecting means; determining means for determining the magnitude of the predetermined value and the identification accuracy based on the area value and the value defining the surrounding contour corresponding to the type of the teaching work; and a value defining the area value and the surrounding contour When the work identification accuracy is high, a value defining the area value and the surrounding contour is selected as the work identification feature value, and when the work identification accuracy is low, the feature value with the highest identification accuracy is used as the work identification feature. A second selection means for selecting as a value, an upper and lower threshold generation means for obtaining upper and lower thresholds based on an average value and a variance of the characteristic values selected by the second selection means, And a target work identifying means for identifying a target work based on the lower threshold.

However, as for the selection of the area value and the value defining the surrounding contour by the second selecting means, a special selecting operation is performed since these characteristic values have already been selected by the first selecting means. Instead, the area value selected by the first selecting means and the value defining the surrounding contour may be used as they are.

[0009]

According to the work identification method of the first aspect, a feature value is extracted based on an image or the like captured by an image detection device using a plurality of teaching works. An area value and a value defining the surrounding contour are selected, and it is determined whether or not each of the workpieces has high identification accuracy. That is, for example, it is determined whether or not the variance ratio of the area value corresponding to the type of the teaching work and the value defining the surrounding contour is larger than a predetermined value. When the work identification accuracy is high, the area value and the value defining the surrounding contour are selected as work identification feature values. Conversely, when the work identification accuracy is low, the work accuracy is maximum, that is, for example, the variance is small. The feature value having the maximum ratio is selected as the work identification feature value. Thereafter, upper and lower thresholds are obtained based on the average value and variance of the selected characteristic values, and the target work can be identified based on the selected work identification characteristic value and the upper and lower thresholds.

According to the work identification device of the second aspect, a feature value is extracted by a feature value extracting means based on an image or the like captured by the image detecting device using a plurality of teaching works. The area value and the value defining the surrounding contour are selected by the selection means, and the determination means determines whether or not the accuracy of the work identification by each is high. Specifically, for example, it is determined whether or not the variance ratio of the area value corresponding to the type of the teaching work and the value defining the surrounding contour is larger than a predetermined value. When the discrimination means determines that the work identification accuracy is high, the second selection means selects an area value and a value defining the surrounding contour as work identification feature values. When it is determined that the characteristic value is low, the second selection unit selects the characteristic value having the maximum identification accuracy, that is, for example, the characteristic value having the maximum dispersion ratio as the work identification characteristic value. Thereafter, upper and lower thresholds are obtained by upper and lower threshold generation means based on the average value and variance of the selected feature values, and the target work is identified by the target work identification means based on the selected work identification feature value and upper and lower thresholds. Can be identified.

More specifically, the object to be identified according to the present invention is a work, and the overall shape and the like are generally considerably different from those for identifying various characters and figures such as characters and figures. For this reason, the present inventor has found that, by preferentially selecting the area value and the value defining the surrounding contour as the feature value, it is often possible to sufficiently increase the work identification accuracy. That is, in such a case, the work can be identified with high accuracy without considering other characteristic values at all, and the load for work identification can be greatly reduced. In addition, when some of the workpieces are to be identified, sufficient identification accuracy cannot be achieved with the area value and the value defining the surrounding contour. In this case, however, the other feature values may not be obtained. Sufficient identification accuracy can be achieved by selecting the object that can achieve the highest identification accuracy from among them and identifying the work based on the selected characteristic value.

As a result, with respect to all the works that can be identified, it is possible to achieve sufficient work identification accuracy and to largely reduce the load for work identification as a whole.

[0013]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. FIG. 7 is a diagram schematically showing the configuration of a component supply system to which the work identification method of the present invention is applied, in which a part of the component transport track 11a of the bowl feeder 11 is formed of a translucent plate 11b. Lighting devices 12 above and below the translucent plate 11b, respectively.
a and 12b are arranged. And the plate 1 with translucency
1b and a lower mirror 12b between the lower lighting device 12b.
c, and the image detection unit 13 is arranged so that light from the translucent plate 11b is reflected by the half mirror 12c and received by the image detection unit 13 including a CCD camera or the like. Further, a suction head 14a for gripping a work is mounted on the industrial robot 14. Also,
An image processing unit 13a that performs a binarization process or the like by using an image signal captured from the image detection unit 13 as an input, and a robot controller that receives a processing result signal from the image processing unit 13a and provides an operation command to the industrial robot 14. 14b
And

FIGS. 1 to 3 are flowcharts showing one embodiment of the work identification method of the present invention. FIG. 1 shows a teaching process of a work to be identified, FIG. 2 shows a feature value selection process for work identification, and FIG. FIG. 3 shows a work identification process based on the selected feature value. In the flowchart of FIG. 1, in step SP11, images of the same type of work having n samples are captured by an image detecting device such as a CCD camera, and in step SP12, an area value and a circumscribed rectangle length are selected as characteristic values. S
In P13, each variance (for example, standard deviation σ)
And an average value xav is calculated, and in step SP14, THup = xav + kσ and THdn = xav-k
The calculation of σ (where k is a constant of about 1 to 3) is performed to calculate the area value, the upper limit threshold THup and the lower limit threshold THdn of the circumscribed rectangle length.

[0015] Step SP11 to step SP1
A series of processes up to 4 are performed for each type of work. In the flowchart of FIG. 2, in step SP21, the variance ratio F0 of each feature value is calculated for the two types of workpieces, and in step SP22, it is determined whether the variance ratio of the area value and the variance ratio of the circumscribed rectangle are smaller than predetermined values. If it is determined that both variance ratios are smaller than the predetermined value, the feature value having the maximum variance ratio is selected in step SP23, and the variance and average value of the selected feature values are calculated in step SP24. In step SP25, as in step SP15 of FIG.
And the lower threshold THdn. Conversely, step S
If it is determined in P22 that the variance ratio is not smaller than the predetermined value, the area value and the circumscribed rectangle length may be used as the characteristic values.
Therefore, the processing of step SP25 is omitted.

In the flowchart of FIG. 3, in step SP31, an image of a work to be identified is fetched, and in step SP32, the characteristic value selected by the processing of FIG. 2 is obtained. Then, in step SP33, it is determined whether or not the area value is between the upper threshold and the lower threshold, and in step SP34, whether or not the circumscribed rectangle length is a value between the upper threshold and the lower threshold. If it is determined in both steps SP33 and SP34 that the value is between the upper limit threshold and the lower limit threshold, the upper limit threshold and the lower limit threshold are calculated in step SP35 in addition to the area and the length of the circumscribed rectangle. It is determined whether or not there is a feature value present, and if it is determined that there is a feature value, whether or not the corresponding feature value is a value between the calculated upper threshold and lower threshold in step SP36 Is determined.

If it is determined in step SP35 that there is no characteristic value, or if it is determined in step SP36 that the characteristic value is a value between the calculated upper threshold and lower threshold, Step SP
At 37, the corresponding work is identified as the work to be processed. Conversely, step SP33, step SP34
Alternatively, if it is determined in step SP36 that the value is not between the upper threshold and the lower threshold, the process proceeds to step SP3
In step 8, the corresponding work is identified as a disturbance work.

More specifically, when the distribution shown in FIG. 4A is obtained for each of the work A and the work B as a result of performing the processing based on the flowchart of FIG. 1, both distributions overlap. Since no part exists, for example, for work A, as shown in FIG. 5, the upper threshold THup and the lower threshold THd
By setting n and determining whether or not the characteristic value obtained for the work to be processed in the flowchart of FIG. 3 is within the range of the upper threshold value THup and the lower threshold value THdn, highly accurate workpiece identification can be achieved. Conversely, FIG.
4B are obtained for the work A and the work B as a result of the processing based on the flowchart of FIG. 4 (the variance ratio of FIG.
If the variance ratio of (B) is FB and FA> FB), there is a considerable overlap between the two distributions. Therefore, even if the processing in the flowchart of FIG. I can't achieve it. Therefore, in this case, by performing the processing of the flowchart of FIG. 2, by selecting the feature value having the largest dispersion ratio, highly accurate workpiece identification can be achieved.

That is, when a work is to be identified,
Unlike characters, figures, and the like, it is often possible to achieve high-precision work identification based on the area value and the circumscribed rectangle length. Therefore, there is no need to select another feature value based on the flowchart of FIG. Since it is only necessary to select another characteristic value based on the flowchart of FIG. 2 for only some of the workpieces, it is possible to reduce the processing load as a whole and achieve highly accurate workpiece identification. For example, when there is only one type of work, it is possible to omit the processing based on the flowchart of FIG. 2 and achieve the identification of a disturbance work in which two or more works overlap.

When there are three or more types of workpieces, the processing of the flowchart of FIG. 2 may be performed for each workpiece. Instead of performing the determination based on the dispersion ratio in step SP22 of the flowchart of FIG. Within the range of the upper threshold value THup and the lower threshold value THdn for one of the workpieces.
Alternatively, it is possible to determine whether or not the lower threshold value THdn exists, and it is also possible to use a circumscribed rectangle width, an inertia ellipse major axis length, or an inertia ellipse minor axis length instead of the circumscribed rectangle length. In addition, various design changes can be made without departing from the scope of the present invention.

[0021]

Embodiment 2 FIG. 6 is a block diagram showing an embodiment of a work identification apparatus according to the present invention, wherein an image detecting section 1 for taking in an image of a work and image taking-in by the image detecting section 1 are repeated a predetermined number of times. A first iteration control unit 2, a feature value extraction unit 1a for extracting a feature value based on a captured image, and selecting an area value and a circumscribed rectangle length from among the extracted feature values for all works. And standard deviation σ and average value xa
v, and THu based on the standard deviation σ and the average value xav calculated by the first calculation unit 3.
a first threshold calculator 4 for calculating p = xav + kσ and THdn = xav-kσ (where k is a constant of about 1 to 3) to calculate an upper threshold THup and a lower threshold THdn, and an area value and a circumscribed rectangle length Value determining unit 5 for determining whether the characteristic value is appropriate as a characteristic value, and the variance ratio is maximum on condition that the area value and the length of the circumscribed rectangle are determined to be inappropriate as characteristic values. A feature value selecting unit 6 for selecting a feature value, and a second calculating unit 7 for calculating a standard deviation σ and an average value xav for the selected feature value.
And THup = xav + kσ and T based on the standard deviation σ and the average value xav calculated by the second calculation unit 7.
Hdn = xav-kσ is calculated and the upper threshold THu is calculated.
Second threshold value calculation unit 8 for calculating p and lower threshold value THdn
A feature value output unit 9 for obtaining and outputting an area value, a circumscribed rectangle length, and a selected feature value if applicable based on the image captured by the image detection unit 1, It is a work to be processed by judging the magnitude of the upper threshold and the lower threshold corresponding to the corresponding feature value among the upper threshold and the lower threshold calculated by the first threshold calculator 4 and the second threshold calculator 8. And a work discriminator 10 that outputs a signal indicating that the work is a disturbance work.

The characteristic value discriminating section 5 discriminates the magnitude of the variance ratio and the predetermined value to determine whether the area value and the circumscribed rectangle length are appropriate as the characteristic values.
It may determine whether or not at least one of the upper threshold and the lower threshold of one of the workpieces is located within the range between the upper threshold and the lower threshold of the other workpiece, and may determine based on other values. May be performed.

The first and second calculation sections 3 and 7 are also used, and the first and second threshold value calculation sections 4 and 8 are used.
Needless to say, both may be used. The operation of the work identification device having the above configuration is as follows. First, an image of the work of the number n of samples is taken in by the image detecting unit 1 for each type of work, and the standard deviation σ and the average value of the area value and the circumscribed rectangle length for each type of work are obtained by the first calculating unit 3. xav is calculated, and based on the calculated standard deviation σ and the average value xav, the first threshold value calculation unit 4 sets THup = xav + kσ and THdn = xav
The calculation of −kσ is performed to calculate the upper threshold THup and the lower threshold THdn. Then, the characteristic value determining unit 5 determines whether the area value and the circumscribed rectangle length are appropriate as the characteristic values.

When the characteristic value discriminating unit 5 determines that the area value and the circumscribed rectangle length are appropriate as the characteristic values, the image of the work to be identified is captured by the image detecting unit 1, and the characteristic value output unit 9. The area value and the circumscribed rectangle length are obtained and output, and the work discriminating unit 10 calculates the magnitude of the characteristic value output from the characteristic value output unit 9 and the upper and lower thresholds calculated by the first threshold calculator 4. A signal indicating that it is a work to be processed or a disturbance work is output. That is, when both output feature values are within the range of the corresponding upper threshold value and lower limit threshold value, a signal indicating that the work to be identified is the work to be processed is output, and at least one feature value is within the above range. If it is outside, a signal indicating that the identification target work is a disturbance work is output.

Conversely, when the characteristic value discriminating section 5 determines that the area value and the circumscribed rectangle length are not appropriate as the characteristic values, the characteristic value selecting section 6 selects the characteristic value having the maximum dispersion ratio. , The second calculator 7 calculates the standard deviation σ and the average value xav of the selected feature value,
THup = xa based on the standard deviation σ and the average value xav calculated by the second calculating unit 7 by the threshold calculating unit 8
The calculation of v + kσ and THdn = xav−kσ is performed to calculate the upper threshold THup and the lower threshold THdn. Thereafter, the image of the work to be identified is captured by the image detection unit 1, the area value, the circumscribed rectangle length, and the value of the feature value selected by the feature value selection unit 6 are obtained and output by the feature value output unit 9. In the unit 10, the magnitude of the characteristic value output from the characteristic value output unit 9 and the upper and lower thresholds calculated by the first threshold calculator 4 and the upper and lower thresholds calculated by the second threshold calculator 8. And outputs a signal indicating that the work is a processing target work or a signal indicating that the work is a disturbance work. That is, when all the output feature values are within the range of the corresponding upper threshold and lower threshold, respectively, a signal indicating that the work to be identified is the work to be processed is output, and at least one feature value is equal to or smaller than the above. When it is out of the range, a signal indicating that the identification target work is a disturbance work is output.

More specifically, when the distribution shown in FIG. 4A is obtained for each of the work A and the work B based on the work of the number n of the samples, there is no overlap between the two distributions. Since they do not exist, for example, the upper limit threshold THup and the lower limit threshold THdn are set for the work (A) with respect to the area value and the circumscribed rectangle length as shown in FIG. The value is the upper threshold THup
By determining whether or not the value is within the range of the lower threshold value THdn, highly accurate workpiece identification can be achieved. vice versa,
When the distribution shown in FIG. 4B is obtained for each of the work A and the work B, since there is a considerable overlap between the two distributions, highly accurate work identification cannot be achieved as it is. Therefore, in this case, highly accurate workpiece identification can be achieved by selecting the feature value having the largest dispersion ratio.

That is, when a work is to be identified,
Unlike characters, figures, etc., it is often possible to achieve high-precision work identification based on the area value and the circumscribed rectangle length, so there is no need to select feature values other than the area value and the circumscribed rectangle length Since it is only necessary to select a feature value other than the area value and the circumscribed rectangle length for only a part of the work, the processing load can be reduced as a whole, and highly accurate work identification can be achieved.

In the first calculating section 3, it is possible to use a width of a circumscribed rectangle, a major axis length of an inertia ellipse, or a minor axis length of an inertia ellipse instead of the circumscribed rectangle length. Various design changes can be made within a range not changed.

[0029]

As described above, according to the first aspect of the present invention, it is possible to achieve highly accurate workpiece identification for various types of workpieces that can be identified, and to significantly reduce the processing load for workpiece identification. To play. The invention of claim 2 also has a unique effect that high-precision work identification can be achieved for various types of works that can be identified, and the processing load for work identification can be significantly reduced.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a process of teaching a work to be identified in the work identification method of the present invention.

FIG. 2 is a flowchart showing a feature value selection process for work identification in the work identification method of the present invention.

FIG. 3 is a flowchart showing a work identification process based on a selected characteristic value in the work identification method of the present invention.

FIG. 4 is a diagram schematically showing the suitability of a feature value;

FIG. 5 is a schematic diagram showing an upper threshold and a lower threshold of a feature value.

FIG. 6 is a block diagram showing one embodiment of a work identification device of the present invention.

FIG. 7 is a diagram schematically showing a configuration of a component supply system to which the work identification method of the present invention is applied.

[Description of Signs] 1a Feature value extraction unit 3 First calculation unit 4 First threshold value calculation unit 5 Feature value determination unit 6 Feature value selection unit 7 Second calculation unit 8 Second threshold value calculation unit 9 Feature value output unit 10 Work determination Department

──────────────────────────────────────────────────続 き Continuation of the front page (51) Int.Cl. ⁶ Identification number Agency reference number FI Technical display location G01B 11/00 G01B 11/00 Z 11/26 11/26 H

Claims (2)

(57) [Claims] 1. A feature value of a plurality of teaching works is extracted, and an area value and a value defining a surrounding contour are selected from the extracted feature values, and an area value corresponding to a type of the teaching work is selected. When the accuracy of identifying the workpiece by the value defining the surrounding contour is high, the area value and the value defining the surrounding contour are selected as the feature values for identifying the workpiece, and when the identifying accuracy of the workpiece is low, the identifying accuracy is maximized. Select a certain feature value as a work identification feature value, obtain upper and lower thresholds based on the average value and variance of the selected feature values, and determine the target work based on the selected work identification feature value and the upper and lower thresholds. A work identification method characterized by identifying. 2. A feature value extracting means (1a) for extracting feature values of a plurality of teaching works;
First selecting means (3) for selecting a value defining an area value and a surrounding contour; and determining the magnitude of a discrimination accuracy between the area value and the value defining the surrounding contour corresponding to the type of the teaching work and the predetermined value. A determination means (4, 5) for determining and a value defining the area value and the surrounding contour when the accuracy of identifying the workpiece by the value defining the area value and the surrounding contour is high; A second selecting unit (6) for selecting a feature value having the highest identifying accuracy as the feature value for identifying the workpiece when the identifying accuracy of the workpiece is low, and an average value of the feature values selected by the second selecting unit (6); Upper and lower threshold generation means (7, 8) for obtaining upper and lower thresholds based on variance
And a target work identification means (9, 10) for identifying a target work based on the selected work identification feature value and upper and lower thresholds.