JP3242530B2 - Work identification method - 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 for identifying the type of a work such as an aluminum wheel or a tire based on its surface shape pattern.

[0002]

2. Description of the Related Art For example, an aluminum wheel is generally provided with a plurality of design holes in response to a demand for appearance or the like. The shape and arrangement of the design holes differ for each type of aluminum wheel. Therefore, by photographing the surface shape of the aluminum wheel, extracting shape data from the image obtained by the photographing, and comparing the shape data with known reference data, it is possible to know what kind of aluminum wheel was photographed. Can be. Examples of the shape data that can be used at that time include the outer diameter of an aluminum wheel, the number of design holes, the area, the pitch, and the like (see Japanese Unexamined Patent Publication No.
225568).

[0003]

However, there are cases where the type of aluminum wheel cannot be identified only by using this kind of shape data. For example, the aluminum wheel 30B shown in FIG. 5 (a) and the aluminum wheel 30C shown in FIG. 5 (b) have almost the same outer shape, the number of design holes, the area, and the pitch. Difficult or impossible.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and has a similar shape by introducing a new index for identifying a work such as an aluminum wheel. An object of the present invention is to enable discrimination between work types that could not be discriminated. Also,
An object of the present invention is to perform this identification more accurately and more easily.

[0005]

In order to achieve the above object, a work identification method according to the present invention comprises the steps of: photographing a work having a surface shape pattern including an element shape; Setting a plurality of detection lines so as to intersect with the element shape, extracting an image pattern of the element shape along each detection line, and comparing the extracted image patterns along the detection lines with each other with the <br/> be, a step of calculating a feature amount having a surface shape pattern of the work, based on the feature quantity obtained, identifying a plurality of kinds of workpieces with different surface shape pattern, the It is characterized by the following.

[0006] The present invention also sets one of the front surface shape Jopa coater <br/> emissions of more reference patterns according to the image pattern of element shapes that are extracted along the plurality of detection lines And calculating a feature amount by comparing an evaluation amount for each image pattern obtained by the correlation between the detection lines with each other, and obtaining a correlation between each of the extracted image patterns and the reference pattern. It is characterized by having.

According to the present invention, an image pattern of an element shape to be obtained when a plurality of types of works are photographed and extracted along an intersection detection line intersecting a detection line is set in advance as an intersection reference pattern. Every time, the image pattern of the element shape is extracted along the intersection detection line, and the correlation between the image pattern of the element shape extracted along the intersection detection line and each cross reference pattern is obtained. Is a target for extracting the image pattern of the element shape along the detection line.

The present invention is characterized in that the characteristic amount is a phase difference of each image pattern.

The present invention is characterized in that, prior to extracting an image pattern of an element shape along a detection line, works having different element shapes or their arrangement orders are excluded from the objects to be extracted.

[0010]

In the present invention, a surface shape pattern of a work is first photographed, and a plurality of detection lines are set so as to intersect with the element shape on the obtained image. When the detection line is set, the image pattern of along element shapes to each detection line is extracted, the feature amount of the surface shape pattern is calculated by a plurality of comparison of image patterns mutual extracted along each detection line for the same work You. That is, when considering a plurality of types of workpieces having the same element shape and the arrangement order and different feature amounts from each other, the difference in the feature amount of the element shape in each type is determined by the number of positions where the detection line intersects with the element shape. Appearing as a relationship between the detection lines, the feature amount of the element shape can be known by focusing on this relationship. The feature amount obtained in this way can be used for identifying the type of work. Therefore, in the present invention, it becomes possible to distinguish between similar works which could not be distinguished conventionally.

In the present invention, a reference pattern is set in advance. This reference pattern is an image pattern of an element shape to be obtained when one of a plurality of types of works to be identified is photographed and extracted along the plurality of detection lines. Further, a correlation between an image pattern extracted along each detection line of the work to be photographed and the reference pattern is obtained. The result of comparing the obtained correlation between the plurality of detection lines on the work related to the photographing indicates how much the work related to the photographing matches or is similar to the type related to the reference pattern when viewed in the image pattern on the detection line. , That is, how similar the positional relationship between the detection line and the element shape is. Therefore, the type of the work can be identified using the result of the comparison. In addition, since the influence of the illuminance and the like at the time of photographing can be eliminated by the correlation calculation, the type of the work can be identified more accurately.

In the present invention, a plurality of types of cross reference patterns are set prior to extracting an image pattern of an element shape along a detection line. Each cross reference pattern is an image pattern of an element shape to be obtained when one of a plurality of types of works to be identified is photographed and extracted along a cross detection line. Also, the intersection detection line
It is set to cross the detection line. With respect to the work related to photography, an image pattern of an element shape is extracted along the intersection detection line, and a correlation operation between the image pattern and each intersection reference pattern is performed. When the obtained correlation is equal to or more than a predetermined value, it is possible to regard that the work related to the photographing and the type related to the cross reference pattern are similar at least in the feature on the cross detection line, and do not exceed the predetermined value. Can be considered dissimilar. Therefore, the frequency of processing such as the extraction of an image pattern along a detection line is reduced by targeting only the work for which the obtained correlation is equal to or greater than a predetermined level to extract an element-shaped image pattern along the detection line. The identification can be realized by simple processing.

In the present invention, the type of the work is identified based on the phase difference of each image pattern. In particular, when there are only a few types of workpieces to be identified, the use of the codes makes the processing quick and simple.

In the present invention, prior to extracting an image pattern of an element shape along a detection line, a work having a different element shape or an arrangement order of the element pattern is identified by, for example, identification processing using conventionally known shape data. Will be excluded from the target. Thereby, the frequency of identification using the index according to the feature of the present invention is suppressed, and the processing load is reduced.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows an environment for implementing a work identification method according to an embodiment of the present invention. As shown in this figure, the present embodiment has an image processing apparatus 1, a CCD camera 2
0, a monitor 21 and a lighting device 22. The lighting device 22 illuminates an aluminum wheel 30 which is a work to be identified,
The CCD camera 20 photographs the aluminum wheel 30. The image processing apparatus 1 inputs an image captured by the CCD camera 20 as an image signal, performs predetermined processing based on the input image signal, identifies the type of the aluminum wheel 30, and displays the result on the screen of the monitor 21. Output to
On the screen of the monitor 21, an image obtained by photographing the aluminum wheel 30, an image obtained in the course of the processing of the image processing device 1, and the like are displayed as necessary.

FIG. 2 shows a functional configuration of the image processing apparatus 1 in this embodiment. In the configuration shown in this figure, the A / D converter 2 is means for converting an image signal obtained by the CCD camera 20 into digital image data, and the image data storage 3 is means for storing this image data. is there. The shape data measurement unit 4, the shape data comparison unit 6, and the shape data storage unit 11
This is a means for deriving shape data of the aluminum wheel 30 photographed from the image data stored above, and identifying the type of the aluminum wheel 30 based on the derived shape data. Shading pattern extracting unit 5, cross-correlation calculating unit 7, shading pattern storing unit 12, phase difference calculating unit 8, and phase difference storing unit 1.
Reference numeral 3 denotes a gray pattern extracted from the image data in the image data storage unit 3 along a predetermined detection line, a cross-correlation coefficient calculated for the extracted gray pattern, and an aluminum wheel based on the obtained cross-correlation coefficient. This is a means for identifying 30 types. The wheel type discriminating unit 9 is a unit that combines the result of the process based on the shape data and the result of the process based on the shading pattern, and outputs the type of the aluminum wheel 30 obtained as a result, for example, on the screen of the monitor 21. Each of the shape data storage unit 11, the light and shade pattern storage unit 12, and the phase difference storage unit 13 is a unit for storing information on a plurality of types of aluminum wheels to be identified, and constitutes the wheel data storage unit 10. ing.

Next, the function of the image processing apparatus 1 will be described with reference to the flowchart shown in FIG.

As shown in FIG. 1, first, the image processing apparatus 1 inputs a captured image of the aluminum wheel 30 as an image signal, converts the image into digital data by the A / D converter 2, and then converts the image data into digital data. It is stored in the storage unit 3 (step 1). The shape data measuring unit 4 extracts a contour image from the image data stored in the image data storage unit 3 (step 2), and obtains the center of the aluminum wheel 30 based on the extracted contour (step 3). For example, when the aluminum wheel 30 being photographed is an aluminum wheel 30A having a shape as shown in FIG. 4A, the wheel outer periphery 31 is detected by the processing of Step 2 and the hub hole is detected by the processing of Step 3. 32 center O
Is detected. In the figure, reference numeral 33 denotes a mounting hole for mounting the aluminum wheel 30;
Is a design hole formed on the surface.

The shape data measuring unit 4 calculates the shape data (wheel outer circumference 31) and the center O based on the thus obtained contour.
Shape data such as the outer diameter of the aluminum wheel 30, the number, area, and pitch of the design holes 34 are obtained (step 4). The shape data comparing unit 6 compares each shape data obtained by the shape data measuring unit 4 with shape data stored as reference data in the shape data storage unit 11 (Step 5). The shape data storage unit 11 stores shape data for each of the multiple processing aluminum wheels to be identified. The shape data comparison unit 6 compares the shape data obtained by the shape data measurement unit 4 with each type of shape data stored in the shape data storage unit 11 one by one, thereby obtaining the shape data obtained by the shape data measurement unit 4. It is checked whether the shape data corresponding to the shape data is stored in the shape data storage unit 11. as a result,
If it is determined that the corresponding shape data is not stored in the shape data storage unit 11, the shape data comparison unit 6
Informs the wheel type discriminating section 9 of this (step 6). In response to this notification, the wheel type discriminating unit 9 considers that the data corresponding to the photographed aluminum wheel 30 does not exist in the wheel data storage unit 10 (step 16), and indicates that fact on the screen of the monitor 21, for example. (Step 15).

When it is determined that the shape data corresponding to the shape data obtained by the shape data measuring unit 4 exists in the shape data storage unit 11, the shape data comparing unit 6 determines the shape data corresponding to the shape data. It is determined whether the type of wheel is one or two or more (step 7). If it is determined that there is only one type, the shape data comparison unit 6 notifies the wheel type determination unit 9 of that fact. In response to the notification, the wheel type determination unit 9 determines the type of the taken aluminum wheel 30 (Step 14), and outputs the result on, for example, the screen of the monitor 21 (Step 15).

If the shape data comparing section 6 determines that the type of the aluminum wheel corresponding to the corresponding shape data is two or more as a result of the determination in step 7, the process proceeds to step 8
Execute the following processing. In step 8, first,
The gradation pattern in the radial direction is extracted from the image data in the image data storage unit 3 by the gradation pattern extraction unit 5. For example, in the example shown in FIG.
A shade pattern is extracted along a broken line 37 connecting the center O of A and the center of gravity G of the design hole 34. On the other hand, on the density pattern storage unit 12, the light and shade pattern along the dashed line 37, is stored as a reference pattern for each type of aluminum foil 30 to be identified. The cross-correlation calculation unit 7
In the following step 9, the light and dark patterns along the broken line 37 stored in the light and dark pattern storage unit 12 as reference patterns are sequentially read out and compared with the light and dark patterns along the broken line 37 extracted by the light and dark pattern extracting unit 5. . That is, the cross-correlation coefficient R (j) of both is obtained. Here, j is a number assigned to the N pieces of data constituting the image pattern, and the cross-correlation coefficient R (j) can be calculated by the following equation.

[0023]

(Equation 1) The cross-correlation coefficient R (j) obtained in this manner is obtained by comparing the gray-scale pattern on the broken line 37 obtained from the image data storage unit 3 with the radial reference pattern stored in the gray-scale pattern storage unit 12. The degree of similarity is shown. That is, when the obtained cross-correlation coefficient R (j) indicates a good correlation, the aluminum wheel 30 stored as image data in the image data storage unit 3 corresponds to the aluminum wheel according to the reference pattern. Can be considered possible. Since the broken line 37 is uniquely determined, the calculation of the cross-correlation coefficient R (j) may be performed only when j = 0 or only in the vicinity of j = 0 (the wheel center O is set to j = 0, J = 0 in the direction away from the center O,
.., N−1). The cross-correlation calculator 7 narrows down the types of the aluminum wheels 30 to be identified in the subsequent processing based on the cross-correlation coefficient R (j).

Next, the gradation pattern extraction unit 5 extracts gradation patterns on concentric circles 35 and 36 centered on the center O of the wheel from the image data stored in the image data storage unit 3 (step 10). . As shown in FIG. 4B, 35 of the broken lines used at this time is a circle having a radius r1 centered on the center O of the aluminum wheel 30A, and a broken line 36 is a circle having a radius r2. In the case of the aluminum wheel 30A shown in FIG. 4A and FIG.
A pattern showing a change in density gradation as shown in FIG.

That is, as shown in FIGS. 4A and 4B, in the case of the aluminum wheel 30A in which the longitudinal direction of the design hole 34 substantially coincides with the broken line indicating the radial direction,
Density gradation change of the shading pattern on the broken line 35;
The above density gradation change occurs at almost the same angle when viewed along the angle θ. On the other hand, as shown in FIG. 5A, the aluminum wheel 30 in which the design hole 34 is inclined is shown.
B, the change in density gradation on the broken line 35 and the broken line 3
As shown in FIG. 7A, a phase difference occurs between the changes in the density gradation on the line 6. The phase difference is obtained by storing in advance the image pattern obtained on the broken lines 35 and 36 for any of a plurality of types of aluminum wheels to be identified as a reference pattern in the gray-scale pattern storage unit 12, By calculating the cross-correlation coefficient with the gray-scale pattern extracted by the gray-scale pattern extraction unit 5, the detection can be performed accurately, that is, without being affected by the illuminance change of the aluminum wheel 30. Therefore,
The cross-correlation calculation unit 7 extracts two types of gray patterns in step 10 and then stores the two types of reference patterns stored in the gray pattern storage unit 12, for example, with respect to FIGS. 4 (a) and 4 (b). The gray-scale pattern is extracted, and the cross-correlation coefficient between the gray-scale pattern obtained by the gray-scale pattern extraction unit 5 and the gray-scale pattern read from the gray-scale pattern storage unit 12 is calculated in accordance with the above-described equation (step 11).

The phase difference calculator 8 calculates the 2
The type of cross-correlation coefficient, that is, the cross-correlation coefficient for the shading pattern on the broken line 35 and the cross-correlation coefficient for the shading pattern on the broken line 36 are compared. Specifically, it is determined how much phase difference the angle at which both cross-correlation coefficients reach a peak has (Step 12). For example, FIG.
When the light and shade pattern according to (a) and (b) is used as a reference pattern and the aluminum wheel 30A is to be photographed, the angle at which the cross-correlation coefficient reaches a peak, as shown in FIG. It coincides between the broken lines 35 and 36.
On the other hand, when the light and shade pattern of the aluminum wheel 30A shown in FIGS. 4A and 4B is used as a reference pattern and the aluminum wheel 30B shown in FIG. As shown in FIG. 7B, a phase difference α occurs between the peak of the cross-correlation coefficient related to the broken line 35 and the peak of the cross-correlation coefficient related to the broken line. The appearance of the phase difference α differs depending on the inclination of the design hole 34. That is, the aluminum wheel 3 shown in FIG.
0B and the aluminum wheel 50C shown in FIG. 5B, the sign of the phase difference α is opposite.

The phase difference storage unit 13 stores, for each type of aluminum wheel to be identified, the type of aluminum wheel to be identified with respect to the density pattern on the broken lines 35 and 36 stored as a reference pattern in the density pattern storage unit 12. The degree of the phase difference is stored. The wheel type discriminating unit 9 calculates the phase difference calculated by the phase difference calculating unit 8, that is, the photographed aluminum wheel 3 with respect to the reference pattern.
The phase difference α of 0 is compared with the phase difference stored for each type in the phase difference storage unit 13, and the data of the type corresponding to the calculated phase difference α is stored in the phase difference storage unit 13. It is determined whether or not it is stored (step 13). If there is no corresponding type of data, the wheel type determining unit 9 determines that there is no corresponding type (step 1).
6), and the fact is output on the screen of the monitor 21 (step 15). Conversely, if there is a corresponding type, the wheel type determining unit 9 determines the relevant type (Step 14) and outputs the result to the screen of the monitor 21 (Step 15).

Therefore, according to this embodiment, the aluminum wheel 30A shown in FIGS. 4A and 4B and the aluminum wheel 30A shown in FIG.
The aluminum wheel 30B shown in FIG. 5B and the aluminum wheel 30C shown in FIG. That is, such aluminum wheels 30A, 3A
Although 0B and 30C cannot be identified by comparison of conventionally known shape data, by using a shading pattern along dashed lines 35 and 36, they can be identified based on the principle shown in FIGS. It can be performed.
In addition, since the cross-correlation coefficients as shown in FIGS. 6B and 7B are used for this identification, it is possible to eliminate the influence of illuminance and the like at the time of photographing. However, even when the processing relating to the cross-correlation coefficient is not performed, since a phase difference appears between the light and shade patterns as shown in FIGS. 6A and 7A, this is used. Identification can also be performed. That is, a rising portion or a falling portion of the light and shade pattern may be detected from, for example, an inflection point or a differential curve thereof, and the phase difference may be obtained using this.

Further, according to the present embodiment, it is possible to suitably and quickly perform identification regardless of the orientation of the aluminum wheel 30 to be photographed. That is, the aluminum wheels 30A, 3A shown in FIG.
0B and 30C each have four design holes 34.
Are formed at intervals of π / 2. Accordingly, the light and shade patterns obtained by photographing these aluminum wheels 30A, 30B and 30C along the broken lines 35 and 36 are shown in FIG.
(A), as shown in FIG. 7 (a), a pattern in which the gradation changes with a period of π / 2 is obtained, and the cross-correlation coefficient is as shown in FIGS. 6 (b) and 7 (b). A peak is found every π / 2. Therefore, in calculating the cross-correlation coefficient, it is not necessary to rotate the reference pattern by 2π in the θ direction, and it is sufficient to rotate the reference pattern by at most π / 2 at the maximum. According to this embodiment, the aluminum wheels 30A, 30B, 30C
The processing can be speeded up depending on the regularity of the shape. Note that θ ₀ shown in FIG. 6B and FIG.
Is the angle of rotation of the reference pattern, that is, the deviation of the attitude of the aluminum wheel 30 that has been photographed.

In the above description, the aluminum wheel 30 is identified. However, the work to be identified in the present embodiment is not limited to the aluminum wheel. For example, a plurality of types of tires can be identified by using a groove pattern formed on the surface. In this case, the lines for extracting the shading pattern may be, for example, two straight lines intersecting the ground contact surface of the tire, instead of the concentric circles of the broken lines 35 and 36 as in the present embodiment.

[0031]

According to the present invention, an image pattern of an element shape is extracted along a plurality of parallel detection lines set so as to intersect with an element shape, and a plurality of image patterns extracted along each detection line for the same work are extracted. Since the inclination of the element shape with respect to the detection line is detected based on the image pattern, a plurality of types of works having the same element shape and the arrangement order and different inclinations can be suitably identified.

Further, according to the present invention, a reference pattern indicating one of the types is set in advance, and an image pattern extracted along each detection line with respect to a work to be photographed, and the reference pattern And the detected correlation is compared between a plurality of detection lines on the work involved in the shooting to detect the inclination, so that the influence of illuminance and the like at the time of shooting can be eliminated, and the type of work can be identified. Becomes more accurate.

According to the present invention, prior to extracting an element-shaped image pattern along a detection line, a plurality of types of cross reference patterns are set along an intersection detection line intersecting with the detection line, and a work related to photographing is set. Calculates the correlation operation between the image pattern of the element shape extracted along the crossing detection line and each cross reference pattern, and extracts the image pattern of the element shape along the detection line only when the obtained correlation is more than a predetermined degree As a result, processing such as extraction of an image pattern along a detection line can be partially omitted, and identification can be realized by simpler processing.

According to the present invention, the type of the work is identified based on the sign or value of the inclination, so that the processing can be performed quickly and simply.

According to the present invention, prior to extracting an image pattern of an element shape along a detection line, works having different element shapes or their arrangement orders are identified by, for example, identification processing using conventionally known shape data. Since it is excluded from the target of extraction, the frequency of identification using the index according to the feature of the present invention can be suppressed, and the processing load can be reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an implementation environment of a work identification method according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of an image processing apparatus according to the embodiment.

FIG. 3 is a flowchart illustrating a flow of an operation of the image processing apparatus according to the embodiment.

FIG. 4 is a diagram showing an aluminum wheel used for setting an identification target and a reference pattern in this embodiment;
FIG. 4A shows the surface shape of this aluminum wheel.
(B) is a figure which shows the setting of the density pattern detection line in this aluminum wheel, respectively.

FIGS. 5A and 5B are diagrams showing examples of an aluminum wheel that can be identified according to the present embodiment.

6 (a) is a gray pattern extracted along the broken lines 35 and 36 from the image data of the aluminum wheel shown in FIG. 4, and FIG. 6 (b) is a cross-sectional view of the gray pattern and the reference pattern. It is a figure which shows a relationship number, respectively.

7A shows a gray pattern extracted from the image data of the aluminum wheel shown in FIG. 5A, and FIG. 7B shows a cross-correlation coefficient between the gray pattern and a reference pattern. FIG.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 image processing device 4 shape data measurement unit 5 density pattern extraction unit 6 shape data comparison unit 7 cross-correlation calculation unit 8 phase difference calculation unit 9 wheel type discrimination unit 10 wheel data storage unit 11 shape data storage unit 12 density pattern storage unit 13 Phase difference storage unit 20 CCD camera 21 Monitor 22 Illumination device 30, 30A, 30B, 30C Aluminum wheel 31 Wheel outer circumference 34 Design hole 35-37 Line for extracting shade pattern O Wheel center G Center of gravity of design hole r1 Radius of broken line 35 r2 Radius of broken line 36 Phase difference

Continuation of the front page (72) Inventor Hiroshi Moribe 41-41, Yokomichi, Chukaku-cho, Nagakute-cho, Aichi-gun, Aichi Prefecture Inside Toyota Central R & D Laboratories Co., Ltd. 60-152907 (JP, A) JP-A-5-272946 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01B 11/00-11/30 102 G06T 1/00- 9/40

Claims (4)

(57) [Claims] 1. A step of photographing a workpiece having a surface shape pattern including an element shape; and setting a plurality of detection lines on the image obtained by the photographing so as to intersect with the element shape. Extracting the image pattern of the along-edge element shape, and calculating the characteristic amount of the surface shape pattern of the workpiece by comparing the extracted image patterns along the respective detection lines with each other. Identifying a plurality of types of workpieces having different surface shape patterns based on the feature amount.
A click identification method, the step of calculating a feature quantity, setting one of a plurality of types of front surface shape Jopa turn as a reference pattern according to the image pattern of the Ru component shape is extracted along the plurality of detection lines and a step advance, determining a correlation between each image pattern and a reference pattern extracted, the evaluation of each image pattern obtained by the correlation detection
A work identification method, comprising: calculating a feature amount by comparing outgoing lines . 2. A step of photographing a workpiece having a surface shape pattern including an element shape, and setting a plurality of detection lines on the image obtained by the photographing so as to intersect with the element shape. By extracting the image pattern of the along-line element shape and comparing the extracted image pattern along each detection line,
Calculating a feature value of the surface shape pattern of the workpiece; and identifying a plurality of types of workpieces having different surface shape patterns based on the obtained feature value. The image pattern of the element shape to be obtained when extracted along the intersection detection line that intersects the detection line is set in advance as an intersection reference pattern, and the image pattern of the element shape is extracted along the intersection detection line, The correlation between the image pattern of the element shape extracted along the intersection detection line and each cross reference pattern is obtained, and only the work having the obtained correlation that is equal to or more than a predetermined degree is subjected to the extraction of the image pattern of the element shape along the detection line. A work identification method. Wherein said feature quantity is a work identification method according to claim 1 or 2, wherein the phase difference between the image patterns. 4. A workpiece identification method of claims 1 to 3 wherein, prior to extracting the image pattern along the detection line element shapes, excludes element shapes or arrangement order is different workpiece from its target of the extraction A work identification method, characterized in that: