JP2004198403A - Inspection device for painting face - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enhance inspection precision (inspection precision for a protrusion and a recess on a painting surface) on the painting surface. <P>SOLUTION: An irradiation device 3 is constituted of bright portions comprising head parts of a plurality of independent light emitting diodes 6 dispersion-arranged along two-dimensional directions comprising the first direction and the second direction crossed with the first direction, and a dark portion 5 other than the plurality of bright portions, and the painting surface is imaged under the condition where inspection light is emitted from the irradiation device, so as to extract a defect. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

  The present invention relates to a painted surface inspection apparatus for inspecting the surface condition of a painted surface, and more particularly, an irradiation unit for irradiating a painted surface with inspection light having a predetermined brightness distribution, and receiving the inspection light. The present invention relates to a painted surface inspection apparatus that includes an imaging unit that captures an image of a painted surface, and an image processing unit that performs image processing on an image captured by the imaging unit.

For example, as disclosed in Patent Documents 1 and 2, a parallel slit having a predetermined width (3, 2c in FIGS. 1 and 2 of Patent Document 1) is used as an inspection device for inspecting a painted surface of an automobile body. The inspection light passing through the parallel slits is radiated from the irradiating means (2a in FIG. 2 of Patent Document 1) to the painted surface, and the image pickup means (5 in FIG. 1 of Patent Document 1) There is an inspection apparatus that takes an image of a painted surface in an irradiation state and inspects the painted surface from an obtained captured image.
In an inspection apparatus of this type, an image processing unit including a microcomputer or the like is provided on the imaging apparatus main body side. The image processing is performed on an image of a painted surface to be imaged, and there is a possibility that the image is present on the painted surface. An imaging region such as a certain defect is extracted as a region of interest.

JP-A-8-145906 (FIGS. 1 and 2) JP-A-9-126744 (FIGS. 1, 3, 4, and 5)

Defects on the painted surface include convex portions and concave portions on the painted surface.
When detecting such projections and depressions using the above-described inspection light in the form of a stripe, the defect image is reflected as a closed region of an intermediate gradation in a dark stripe existing in the captured image, for example. Put in.

In the techniques disclosed in Patent Literatures 1 and 2, since a stripe-shaped inspection light having a predetermined width is used, a high-luminance portion corresponding to a bright portion on the irradiation side and a low-luminance portion corresponding to a dark portion are used. Is a single straight line.
Thereby, in the transverse direction orthogonal to the stripe, the boundary between the high-luminance portion corresponding to the bright portion and the low-luminance portion corresponding to the dark portion is in a state in which the boundary exists at a predetermined interval, In the vertical direction parallel to the stripe, there is no boundary between a high luminance portion corresponding to a bright portion and a low luminance portion corresponding to a dark portion.

As a result, depending on the defect shape and the like, for example, even if there is a defect equivalent to or less than the stripe width at a position facing the dark stripe, reflection from the bright stripe portion on both sides in the transverse direction of the dark stripe does not occur. In this case, that is, when reflection occurs only in a direction parallel to the stripe, this defect cannot be identified as a light and dark pattern, and a projection or a depression on the painted surface may be overlooked.
An object of the present invention is to provide a coating surface inspection apparatus capable of extremely reducing the probability of this kind of oversight and detecting a relatively small defect even if the defect is relatively small. And

[I]
According to the feature of the first aspect, the light and dark distribution on the irradiation surface of the irradiation means is distributed in a two-dimensional direction consisting of a first direction and a second direction intersecting the first direction. It is composed of a light part and a plurality of dark parts other than the light part,
When a captured image in a state where a normal painted surface is captured is set as a normal captured image, an imaging region corresponding to the bright portion in the normal captured image is a normal bright imaging region, and an imaging region corresponding to the dark portion is normal. As a dark imaging region, a configuration is provided in which a halftone region between the normal bright imaging region and the normal dark imaging region is extracted as a region of interest.

Thus, according to the features of the first aspect, the boundary between the high-brightness portion corresponding to the bright portion and the low-brightness portion corresponding to the dark portion is not a single linear shape but a single closed loop shape. (For example, circular or polygonal), and a boundary between a high-luminance portion corresponding to a bright portion and a low-luminance portion corresponding to a dark portion corresponds to a first direction and a second direction intersecting the first direction. It will be in the state of being arranged.
Therefore, according to the features of claim 1, it can be said that a boundary between a high-luminance portion corresponding to a bright portion and a low-luminance portion corresponding to a dark portion exists in one direction but does not exist in another direction. In this case, the intermediate luminance portion caused by the defect can be surely formed, and a convex portion, a concave portion, and the like existing on the painted surface are not easily overlooked. As a result, the reliability of the inspection is significantly improved.
Further, since the light-dark distribution can be easily formed as a fine distribution, the opportunity of reflection of dark light from a dark portion existing around the bright portion and reflection of bright light from the bright portion existing around the dark portion in the present application is remarkable. As a result, it has become possible to detect even minute defects that have been difficult to detect.

[II]
Further, as described in claim 2, the shapes of the plurality of independent bright portions are all the same, and the bright portions are uniformly distributed in the two-dimensional direction,
In the captured image, a bright imaging region that is an imaging region corresponding to the independent bright portion, and a dark imaging region around the bright imaging region as a unit determination region,
It is preferable to include a discriminating means for discriminating the normality / abnormality of the painted surface corresponding to each unit discrimination area from the imaged light / dark pattern of the unit discrimination area.

  By separately providing a unit determination area and providing a determination unit for determining whether the part is normal or abnormal, a predetermined divided area is individually set within the imaging area, and the area is determined within the area. , A region having a high possibility of having a defect can be extracted first.

[III]
Then, as described in claim 3, after executing the normal / abnormal judgment for each of the unit judgment areas by the judgment means,
By extracting the intermediate gradation area as a region of interest for the unit determination region determined to be abnormal, the region of interest can be quickly extracted and a region likely to be defective is determined based on detailed determination criteria. Can be executed.
In this case, the determination by the determination means and the extraction process are performed twice, so that the reliability is significantly improved.

[IV]
According to the feature of claim 4, the bright portion is set in a circular shape, and the boundary between the high-luminance portion corresponding to the bright portion and the low-luminance portion corresponding to the dark portion is circular. In addition, a state in which the detection characteristics of the convex portion, the concave portion, and the like of the painted surface differ depending on the positions of the corners and the sides of the polygonal shape does not occur.

[V]
According to the feature of claim 5, as the boundary of the light and dark pattern in the captured image received by the imaging means becomes clearer (in a state where bleeding is less and not disturbed), the inspection accuracy of the painted surface (such as a convex portion or a concave portion of the painted surface) Detection accuracy).
That is, by configuring the light portion by the light emitting diode, a bright portion having sufficiently high luminance can be formed, and a high brightness region corresponding to the bright portion and a low brightness corresponding to the dark portion on the imaging side are formed. The boundary between the luminance region and the luminance region becomes clear (in a state where blur is small and is not disturbed), and highly reliable detection suitable for the purpose of the problem can be performed.

[Usage of inspection equipment]
FIG. 1 shows a state in which a painted surface of an automobile body 1 is inspected using the inspection apparatus 100 of the present application. The painted body 1 is conveyed by a conveyor 2 to the left in the drawing. come.

  The inspection device 100 includes an irradiation device 3 and an imaging device 4 as its irradiation / imaging system, and a control unit 7 for controlling the operation of the entire inspection device, and a captured image obtained by the imaging device 3 The image processing apparatus includes an image processing unit 8 for performing image processing, and a monitor unit 10 for displaying a captured image, an image after image processing, and the like.

  As shown in FIG. 1, an irradiation device 3 is disposed on the upper side of the conveyor 2 (on the right side in FIG. 1), and an imaging device 4 (for example, a CCD) is on the lower side of the conveyor 2 (on the left side in FIG. 1). A camera or the like is arranged, and the inspection of the painted surface is executed.

The painted surface inspection apparatus 100 outputs, to the computer 9 provided on the lower side of the information processing, a region of interest extracted as possibly having a painted surface defect as defect information together with its attribute data. The computer 9 separately performs necessary processing such as repainting based on the input defect information.
On the other hand, at present, the configuration is such that the position information of the coating part (inspection part) that is inspected by the inspection apparatus 100 is input to the inspection apparatus 100 side. It is configured such that a captured image to be obtained by the imaging device 3 (called a normal captured image) can be specified.

[Irradiation device and imaging device]
As shown in FIG. 1, the irradiation device 3 and the imaging device 4 are arranged side by side so as to be substantially orthogonal to a direction A0 orthogonal to the transport direction of the conveyor 2.
That is, the irradiation surface 3a of the irradiation device 3 and the imaging surface 4a of the imaging device 4 are disposed substantially parallel to the painted surface. In the same drawing, the irradiation direction is shown inclined to facilitate the display of the drawing.

  As shown in FIG. 2, the irradiation device 3 includes an outer frame 5 and a light emitting diode 6 arranged inside the outer frame 5. The tip shape of the light emitting diode 6 is circular when viewed from the front, and a plurality of light emitting diodes 6 are arranged at predetermined intervals in the vertical direction and in the horizontal direction intersecting the vertical direction. Here, the diameter of the light emitting diode 6 is about 3 to 5 mm. Therefore, as will be clear from the description on the image processing side, the size of the defect that can be detected in this embodiment is about 0.3 mm to 3.0 mm.

  The other outer frame portion 5 is colored black. That is, the light portion, which is the light emitting portion of the light emitting diode 6, is arranged equally in two directions in the vertical and horizontal directions in the dark portion of the blackboard.

  As shown in FIG. 1, the imaging device 4 is configured to image the irradiation surface 3 a of the lighting device 3 (the surface where the head of the light emitting diode 6 is exposed) via the painted surface of the body 1. In the inspection of the painted surface in the present application, the focus of the imaging device 4 is adjusted to the position of the irradiation surface 3a of the irradiation device 3 before use.

  In imaging, the focus is adjusted to the irradiation surface 3a of the irradiation device 3, for example, when an image is taken of a light portion which is a light emitting portion of the light emitting diode 6, a boundary between the light portion and the dark portion is clearly identified. This is because it becomes possible.

[Captured image]
In the painted surface inspection apparatus according to the present invention, on a planar painted surface, as an example, a captured image received by the imaging device 4 is as shown in FIGS.
In this case, a high-luminance K1 region (bright imaging region) C1 corresponding to a bright portion (light-emitting diode 6) and a low-luminance K2 region (dark imaging region) C2 corresponding to a dark portion (outer frame portion 5) are obtained. In a normal portion of the painted surface, a region (normal bright imaging region) C1 of high luminance K1 corresponding to a bright portion has a circular shape as a representative of the shape on the irradiation side.
If the painted surface has a defect such as irregularities, the curved surface shape may affect the region C3 of the high brightness K1 corresponding to the bright portion, and the region C3 may become elliptical instead of circular.
3 and 4 show a case where a defect exists at a substantially middle position between the four distorted bright imaging regions indicated by C3.

Depending on other defects of the painted surface, as shown in FIGS. 5 (a) and (b), in a portion such as a convex portion or a concave portion of the painted surface, a region C4 of high luminance K1 corresponding to a bright portion has a low luminance in the center. 6B or a slightly higher luminance area C5 in the low luminance K2 area C2 corresponding to the dark part as shown in FIGS. Sometimes.
In the example shown in FIGS. 5 (a) and 5 (b), the area that should be a simple bright imaging area is within the bright imaging area from the dark area around the illumination-side bright area due to the influence of the defect shape and the like. , And an intermediate gradation area as shown by C4 is formed.
Such an image is formed when a defect slightly smaller than the size of the tip of the light-emitting diode is present at a position substantially directly opposite the center of the light-emitting diode.

  The example shown in FIGS. 6A and 6B is a case where there is a defect of approximately the same size at a position corresponding to the irradiation side dark portion indicated by C5 in FIG. In this case, reflections from four bright portions existing around the image occur. Therefore, also in this case, the problematic area is an intermediate gradation area.

  As described above, the imaging mode differs in correspondence with a normal painted surface due to a defective portion such as a convex portion or a concave portion of the painted surface. It is considered that the inspection light from the diode 6 is irregularly reflected, and the reflection referred to in the present application occurs.

  In this case, since the focus of the imaging device 4 is set to the position of the irradiation surface 3a of the irradiation device 3, as shown in FIGS. The boundary between the corresponding high luminance K1 area C1 and the low luminance K2 area C2 corresponding to the dark part (outer frame part 5) becomes clear, and the luminance of the high luminance K1 area C1 corresponding to the bright part is reduced. The brightness difference between the high brightness K1 region C1 corresponding to the bright portion (light emitting diode 6) and the low brightness K2 region C2 corresponding to the dark portion (outer frame portion 5) becomes large. ).

  That is, the boundary in the case of distinguishing between light and dark areas can be sharpened. Further, even when the reflection that forms the intermediate gradation area occurs, the boundary of the image area where the reflection occurs can be sharpened. It can be something.

Hereinafter, the image processing in the image processing unit 8 will be described.
In the present application, first and second image processing modes are shown as image processing modes. FIG. 7A shows an image processing procedure of the first image processing mode in the present application, and FIG. 7B shows an image processing procedure of the second image processing mode.
In the first image processing mode, each image processing is executed uniformly and sequentially using the entire captured image as a processing target area. In the second image processing mode, the entire image is corrected by a paint color. After performing the processing, pattern recognition is performed to compare the light and dark pattern around the normal light and dark image area, which is the image area corresponding to the light area obtained when the normal painted surface is imaged, with the light and dark pattern in the actual image. The method is used to determine whether the part is normal or abnormal, and thereafter, the remaining processing of the first image processing mode is executed only for the area determined to be abnormal.

[First image processing mode]
In this image processing mode, as shown in FIG. 7A, after performing a correction process (# 2) based on a paint color, a ternarization process (# 3) and elimination of an image area corresponding to a normal paint surface. The processing (# 4, 5) is performed to specify and extract the final region of interest, perform post-processing (# 6: area determination processing, labeling processing, area / centroid calculation processing), and output to the computer 9 side (# Execute 7).
For this purpose, as shown in FIG. 1, the image processing section 8 is provided with a correction means 8a, a ternarization means 8b, an exclusion means 8c, and a post-processing means 8d based on a paint color.

  Hereinafter, the first image processing will be described with reference to FIGS. In order to facilitate understanding of the processing of the present application, the correction processing by the coating processing will be described after the ternary processing, the elimination processing, and the like are described.

a Processing status when the painted surface is flat A. Ternary processing (processing by the ternary means 8b)
The captured image shown in FIG. 3A is ternarized by the image processing device 8 and corresponds to a region C1 of high luminance K1 corresponding to a bright portion (light emitting diode 6) and a dark portion (outer frame portion 5). It is divided into a region C2 of low luminance K2 and a region C6 of intermediate luminance between high luminance K1 and low luminance K2.

(B) exclusion processing (processing by the exclusion means 8c)
Subsequently, a region C1 of high luminance K1 corresponding to a bright portion (light emitting diode 6) and a region of low luminance K2 corresponding to a dark portion (outer frame portion 5) are removed. As a result, as shown in FIG. 3B, a region C6 of intermediate luminance, a region C1 of high luminance K1 corresponding to a bright portion (light emitting diode 6) and a low luminance corresponding to a dark portion (outer frame portion 5). The boundary C7 between the region K2 and the region K2 of K2 remains.
Further, a so-called expansion / contraction process is performed on the processed image obtained by the above procedure. Since the boundary C7, which is an area to be extracted in the inspection of the painted surface, is smaller than the intermediate gradation area C6, if the expansion and contraction processing is performed several times after the above, as shown in FIG. The above-mentioned boundary C7 disappears, and a region C6 of an intermediate gradation remains.
In this process, the captured image portion where painting is normally performed is substantially removed.
The intermediate gradation area C6 can be identified as a convex portion, a concave portion, or the like of the painted surface of the body 1, and can be set as a region of interest.

C Post-processing (processing by post-processing means 8d)
As a result, the area determination processing, the labeling processing, and the area and center-of-gravity calculation processing are performed on the remaining intermediate luminance area C6.
Here, the area determination is to determine the likelihood of a defect based on a predetermined threshold value in accordance with the area of the region, and the labeling is sequentially performed with respect to the region considered to be a defect. And the like.
Further, the area / center-of-gravity calculation processing is for obtaining the area / center-of-gravity position of each region, and data obtained by these post-processing is output to the computer 9 described above.

b. Processing when the painted surface is a curved surface Even when the painted surface is a curved surface, the image processing procedure is the same as the processing for a flat surface.
With respect to a flat painted surface, as shown in FIG. 8, on a curved painted surface having a vertical axis as a center line, a captured image received by the imaging device 4 is as shown in FIG. It becomes. Here, the reason why the C1 portion is sequentially moving from the upper left direction to the lower right direction is that the painted surface has a predetermined shape in accordance with the design of the body (however, FIGS. 8 and 10). Then, detailed shape display is omitted).
At the time of the image processing, the captured image is subjected to the ternarization processing so that the captured image has a high luminance K1 region C1 corresponding to a bright portion (light emitting diode 6) and a low luminance corresponding to a dark portion (outer frame portion 5). An area C2 of K2 and an area C6 of intermediate luminance between the high luminance K1 and the low luminance K2 are identified.

Further, a region C1 of high luminance K1 corresponding to a bright portion and a region C2 of low luminance K2 corresponding to a dark portion are excluded. As a result, a processed image shown in FIG. 9B can be obtained.
Thereafter, expansion and contraction processes are performed several times to remove the boundary C7 between the high-luminance K1 region C1 corresponding to the bright portion and the low-luminance K2 region C2 corresponding to the dark portion. ), An area C6 of intermediate luminance is left. As a result, it is possible to determine the region C6 of the intermediate luminance as a convex portion, a concave portion, or the like of the painted surface of the body 1.
The remaining intermediate luminance area C6 is set as a target area, area determination processing, labeling processing, area and centroid calculation processing are performed, and the result is output to the computer 9 side.

On the other hand, as shown in FIG. 10, on a curved painted surface whose center line is the axis in the front-rear direction, the captured image received by the imaging device 4 is as shown in FIG.
Accordingly, the image data is processed by the image processing device 8, and the area C1 of the high luminance K1 corresponding to the bright part, the area C2 of the low luminance K2 corresponding to the dark part, and the intermediate between the high luminance K1 and the low luminance K2. The luminance area C6 is separated.

Thereafter, the region C1 of the high luminance K1 corresponding to the bright portion and the region C2 of the low luminance K2 corresponding to the dark portion are removed, and the processed image shown in FIG. 11B can be obtained.
Subsequently, the expansion and contraction processing is performed several times, and the boundary area C7 between the area C1 of the high luminance K1 corresponding to the bright part and the area C2 of the low luminance K2 corresponding to the dark part disappears, and the part of the intermediate luminance disappears. C6 remains, and as shown in FIG. 11C, the portion C6 of the intermediate luminance can be determined as a convex portion, a concave portion, or the like of the painted surface of the body 1.
Thus, the area determination processing, labeling processing, area and center-of-gravity calculation processing are performed on the remaining intermediate luminance area C6, and the result is output to the computer 9.

D. Correction processing based on paint color (processing by correction means 8a)
As described above, the brightness pattern appearing in the captured image depends on the painting color of the body 1 (for example, white or black) and the size of the high-luminance K1 region C1 corresponding to the bright portion and the medium-luminance region C6. Is affected.

  In general, in the paint color of the white body 1, the luminance of the low luminance K2 area C2 corresponding to the dark part is relatively high, and the high luminance K1 area C1 corresponding to the bright part and the low luminance corresponding to the dark part. The luminance difference between the region C2 of K2 and the region C2 becomes relatively small (a state where the difference in shading is small).

  On the other hand, in the paint color of the black body 1, the brightness of the region C2 of the low brightness K2 corresponding to the dark portion (outer frame portion 5) is relatively low, and the high brightness K1 corresponding to the light portion (the light emitting diode 6). And the region C2 of low luminance K2 corresponding to the dark portion (outer frame portion 5) have a relatively large luminance difference (a state in which the density difference is large). Further, the area C1 of the high luminance K1 corresponding to the bright part (the light emitting diode 6) and the area C6 of the intermediate luminance become relatively large.

  By ignoring the effect of this kind of paint color and executing the image processing described so far, the area C1 of the high luminance K1 corresponding to the bright part (light emitting diode 6) and the area C6 of the intermediate luminance are relatively small. The processing is performed while it is still large or large, and good inspection results cannot be obtained.

  Therefore, in the captured image received by the imaging device 4, as shown in FIG. 12, according to the luminance difference between the area C1 of the high luminance K1 corresponding to the bright part and the area C2 of the low luminance K2 corresponding to the dark part. The size of the high-luminance K1 region C1 and the middle-luminance region C6 corresponding to the bright portion is corrected so as to be the reference size (step # 2 in FIG. 7). This correction is based on correction data prepared in advance according to the paint color.

  For example, when the reference paint color is gray, which is an intermediate gradation between white and black, and when the paint color is white, correction is performed based on a state in which the tip of the light emitting diode is reflected as a bright area having a diameter of 1 mm. If there is no bright area with a diameter of 1.2 mm, the image processing side corrects it so that the size of the bright area becomes 1.0 mm. On the other hand, when the paint color is black, if the correction is not performed, the image is reflected as a bright area having a diameter of 0.8 mm. Similarly, correction is performed on the image processing side so that the size of the bright area becomes 1.0 mm.

Thereafter, as shown in FIG. 7A, by performing image processing of the corrected captured image, a favorable inspection result which is not affected by the paint color can be obtained.

[Second image processing form]
In the above-described first image processing mode, after performing substantial enlargement / shrinkage correction of the area by the paint color, the image after the correction is subjected to ternarization processing and exclusion processing (high brightness area, The low-luminance area and the boundary area are excluded), and finally, an image area that does not appear in the captured image on the normal painted surface in the intermediate-luminance area is determined and determined as the attention area. In the image processing of this mode, in each processing, all the regions of the captured image are processed.

However, for example, if a notable area can be extracted at an early stage of image processing, the processing can be advanced quickly. The second image processing mode in the present application is such a mode.
After the problematic area is extracted, ternarization processing and elimination processing (elimination of high-luminance area, low-luminance area, and boundary area) are performed, and finally, normal coating is performed in the middle-luminance area. An image area that does not appear in the captured image on the surface is determined and determined as the attention area. In the initial stage of the image processing, the substantial enlargement / shrinkage correction of the area by the paint color is performed (# 2), and there is no change in this mode.

The image processing in this mode is, as shown in FIG. 7B, a process of enlarging / shrinking a bright / dark area according to the paint color (# 2), and a paint surface based on pattern recognition for each unit determination area. Normal / abnormal determination processing (# 3, 4), ternarization processing (# 6-1) for the unit determination area determined to be abnormal, and exclusion processing (high luminance area, low luminance area, and boundary area Exclusion: # 6-2, # 6-3), identification of the final intermediate luminance area as a target area, post-processing (# 6-4), and output (# 7).
For the above purpose, as shown in FIG. 13 corresponding to FIG. 1, in the inspection apparatus 100 that executes this processing, the image processing unit 8 is provided with a separate determination unit 8e.

Hereinafter, a description will be given of a normal / abnormal determination process of a painted surface based on pattern recognition for each unit determination region, which is a feature of this embodiment.
B. Classification of unit discrimination area (# 3)
As described above, in the present application, due to the configuration of the shape and distribution of the light portion and the dark portion on the irradiation surface 3a of the irradiation device 3, the dark portion is formed around the light portion formed by the tip portion of the light emitting diode 6. They are arranged uniformly distributed in the two-dimensional direction.

Therefore, at the time of image processing, each of a closed imaging region, which is an imaging region corresponding to an independent bright portion, and a region obtained by combining a dark imaging region around the bright imaging region is defined as a unit. Handled as the discrimination area A.
An example of the unit determination area A is an area surrounded by a chain line in FIG.

B. Judgment of normal / abnormal of painted surface by pattern recognition (# 4)
For each of the unit determination areas A determined as described above, a so-called pattern recognition method is used to determine whether there is a high possibility that an image area due to a defect is contained in the unit determination area A. To judge.
At the time of the determination, the light and dark pattern information on the captured image in the unit determination area A in a state where the painting is normally performed on the body having the same vehicle type, painting color, and inspection condition is used.
As shown by the two-dot chain line in FIG. 15A, the light / dark pattern information of the normal painted surface is obtained from the captured image at a plurality of locations in the horizontal direction (each transverse line is indicated by a circled number). This is information stored by capturing light-dark pattern information in the direction. That is, the light and dark pattern information is in the form shown in FIG. Here, for ease of understanding, the number of crossing points is set to 7, but in practice, the number is set to a number that can collect the frequency distribution of the coincidence.

On the other hand, as for the image of the unit discrimination area A to be inspected, similarly to the normal image, the light-dark pattern in that direction is captured at a plurality of positions in the horizontal direction and is compared with the light-dark pattern of the normal painted surface.
Then, as shown in FIG. 16, the frequency distribution of the coincidence is the frequency distribution of the coincidence (OK side) and the coincidence of the frequency distribution (OK side) for all of the plurality of locations in the horizontal direction (for example, all the crossings 1 to 7). It is obtained as a frequency distribution (NG side) of those that do not.

  According to the distribution obtained in this way, as shown in FIG. 16, it is determined that the coating having a high frequency on the matching side is normal, and the frequency on the non-matching side is not shown. If it is high, it is judged that there is a high possibility that the painted surface has a defect (abnormal).

  As a result, the normality / abnormality of the painted surface corresponding to each unit determination area can be determined by pattern recognition of the light and dark patterns.

  Then, the unit determination area A determined to be abnormal in this way can be extracted as a final area of interest in the same manner as in the procedure of the first image processing.

[Another embodiment of the invention]
(1) In the irradiation device 3 of the above-described embodiment shown in FIG. 2, a black board (not shown) is arranged in front of the light emitting diode 6, and a circle is provided on a portion of the black board corresponding to the position of the light emitting diode 6. It may be configured to provide an opening in the shape of a circle.
With this configuration, since the boundary between the light emitting diode 6 and the black board in the irradiation device 4 becomes clear, FIGS. 3 (A), 5 (A), 6 (A), 9 (A), and 11 (A). The boundary between the portion C1 of the high brightness K1 corresponding to the bright portion (light emitting diode 6) and the portion C2 of the low brightness K2 corresponding to the dark portion (outer frame portion 5) becomes clearer.
The arrangement of the light emitting diodes 6 in the irradiation device 3 is not limited to the arrangement as shown in FIG. 2 of the above-described embodiment, but may be arranged radially (Gaussian function) from one center or arranged in a staggered manner. May be.
(2) In the above-described embodiment, the pattern recognition method is used to determine whether each unit determination area A is normal or abnormal. For example, a predetermined number of representative positions may be set in a bright area. May be determined, and the degree of matching of the light and dark intensities at these portions may be checked. Further, a technique such as area comparison may be employed.
In order to check the degree of coincidence between the light and dark patterns as described above, the judgment may be made not only in the transverse direction but also in any direction.
(3) In the above embodiment, an example in which a single bright portion is included in each unit determination area A has been described. However, depending on the processing speed of the image processing apparatus, for example, a 2 × 2 bright portion , And may further include a bright portion.
However, at present, by selecting about 12 × 12, image processing at a processing speed in accordance with the actual situation of the inspection of the painted surface can be realized.
(4) The shape of the bright area on the imaging surface of the imaging apparatus has been described by exemplifying the case of a circle and a polygon. However, any shape of a closed area may be used. A shape that becomes round or polygonal on the side may be selected.

  For example, an inspection of a painted surface of an automobile body can be reliably performed.

The figure which shows the situation which is performing the inspection using the coating surface inspection apparatus which concerns on this application. Perspective view of irradiation device The figure which shows the processing image of the situation which extracted the picked-up image, the image after ternarization processing, and the attention area after dilation / contraction processing FIG. 3 is a three-dimensional representation of a light and dark pattern of the captured image shown in FIG. A three-dimensional representation of a picked-up image in which a reflection from a dark portion around a bright portion occurs due to a defect in a bright imaging region corresponding to a bright portion, and its light-dark pattern A three-dimensional representation of a picked-up image in which a reflection from a bright part around a dark part occurs due to a defect in a bright imaging area corresponding to a dark part and its light-dark pattern. FIG. 3 is a diagram showing a processing flow of a first image processing mode and a second image processing mode according to the present application. A perspective view showing a state of an inspection on a curved painted surface with a vertical axis as a center line. FIG. 8 is a diagram showing a captured image captured by the configuration of FIG. Perspective view showing the state of inspection on a painted surface with a curved surface centered on the longitudinal axis. FIG. 10 is a diagram showing a captured image captured by the configuration of FIG. Diagram showing the state of correction performed according to the paint color of the body The figure which shows the use condition of the inspection device which performs the image processing which finds the unit discrimination area determined as abnormal by the pattern recognition technique. Diagram showing division of unit determination area Diagram showing recognition areas and pattern of light and dark in pattern recognition Diagram showing frequency distribution of coincidence

Explanation of reference numerals

3 irradiating means 4 imaging means 6 light emitting diode

Claims (5)

Irradiating means for irradiating the painted surface with inspection light having a predetermined brightness distribution, imaging means for imaging the painted surface having received the inspection light, and image processing for image-processing an image captured by the imaging means Means for inspecting a painted surface comprising:
A plurality of independent light portions distributed in a two-dimensional direction including a first direction and a second direction intersecting with the first direction, wherein the light / dark distribution on an irradiation surface in the irradiation means is; And dark parts other than the light part of
When a captured image in a state where a normal painted surface is captured is set as a normal captured image, an imaging region corresponding to the bright portion in the normal captured image is a normal bright imaging region, and an imaging region corresponding to the dark portion is normal. A painted surface inspection device for extracting, as a dark imaging region, an intermediate gradation region between the normal bright imaging region and the normal dark imaging region as a region of interest. The shapes of the plurality of independent bright portions are all the same, and the bright portions are uniformly distributed in the two-dimensional direction,
In the captured image, a bright imaging region that is an imaging region corresponding to the independent bright portion, and a dark imaging region around the bright imaging region as a unit determination region,
2. The painted surface inspection apparatus according to claim 1, further comprising a discriminating means for discriminating whether the painted surface corresponding to each unit discrimination region is normal or abnormal based on an imaged light / dark pattern of the unit discrimination region. After executing the normal / abnormal judgment for each unit judgment area by the judgment means,
3. The painted surface inspection apparatus according to claim 2, wherein the halftone area is extracted as the attention area for the unit determination area determined to be abnormal. The painted surface inspection apparatus according to any one of claims 1 to 3, wherein the bright portion is set in a circular shape. The painted surface inspection apparatus according to any one of claims 1 to 4, wherein the bright portion on the irradiation surface is set by a light emitting diode.

Images