JP2008190872A - Surface defectiveness detector, surface defectiveness detecting method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing technique capable of detecting the fine damage group (white blur) of a coating surface, with precision equal to or higher than that of the organoleptic evaluation by a skilled technical person. <P>SOLUTION: The detected image data obtained by irradiating an inspection surface with light and imaging the light reflected from the inspection surface are subjected to addition averaging to form an averaged image, the averaged image is contracted by using a bistable primary interpolation method in a predetermined contraction magnification to form a contracted image; the contacted image is expanded in a predetermined magnification by using the bistable primary interpolation method to form an expanded image; the luminance histogram of the expanded image is formed, the luminance histogram is analyzed to calculate standard deviation; and the standard deviation is compared with a preset threshold to determine the existence of surface defects. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technique for detecting surface defects such as a group of minute scratches on a painted surface.

  2. Description of the Related Art Conventionally, it has been performed to irradiate a painted surface such as a vehicle body with light and analyze the reflected light to detect the presence or absence of surface defects such as scratches and irregularities on the painted surface. For example, the technique described in Patent Document 1.

  In the technique of Patent Document 1, first, light is applied to a painted surface from a light source, and an image of the painted surface is captured by a CCD camera and captured as image data. Next, after performing luminance histogram analysis or line profile analysis on the image data with a computer and calculating a quantitative evaluation value related to metallic coating unevenness, the evaluation value is compared with a preset reference value to determine uneven coating. The presence or absence of occurrence is determined.

  If a surface defect is found on the painted surface as described above, it is repaired by polishing the defective portion. However, secondary surface defects may occur due to a group of minute scratches generated during polishing. This group of minute wounds is usually called “white blur”.

  In recent years, “metallic paint” has become the mainstream for painting vehicle bodies. Metallic paint is a solid color paint mixed with metallic (fine aluminum pieces). A metallic paint layer is formed by laminating multiple undercoat layers on an iron plate, and then a transparent paint for coating the metallic paint A clear layer is formed by painting. In this way, when light is applied to the painted surface of the vehicle body that has been subjected to the metallic coating, the light is reflected by both the clear layer on the surface layer side and the metallic layer below it. In addition to the micro-scratch group, the light reflected also by the metallic contained in the metallic layer.

In the image obtained by irradiating the painted surface including the micro-scratch group with light and imaging the reflected light, the brightness difference occurs in the part where the micro-scratch group exists, so that this brightness difference It is theoretically possible to know the presence or absence of a micro-flaw group by detecting.
However, for example, as shown in FIG. 7, when the standard deviation of the luminance distribution is calculated from image data obtained by imaging the reflected light from the inspection surface coated with metallic paint, the variation in luminance distribution (luminance difference) is metallic. Based on the standard deviation of the luminance distribution, it is difficult to distinguish between the reflection from the (or metallic layer) or the minute wound group, and the reflection from the metallic and the minute wound group may overlap. It is difficult to determine the presence or absence of surface defects by quantitatively evaluating only the micro-flaw group.
Therefore, an optimal image processing method has not been established for accurately detecting a group of minute flaws (white blur) on the metallic paint surface, and it is necessary to rely on sensory evaluation by a skilled engineer. .
JP 2000-205846 A

  In view of the above-described conventional technology, the present invention proposes an image processing technique that can detect a minute flaw group (white blur) on a painted surface with an accuracy equal to or higher than a sensory evaluation by a skilled engineer. To do.

  The problem to be solved by the present invention is as described above. Next, means for solving the problem will be described.

  That is, according to claim 1, addition averaging means for creating an averaged image by averaging the detected image data obtained by irradiating the inspection surface with light and imaging the light reflected by the inspection surface, and bilinear interpolation An image reduction means for creating a reduced image obtained by reducing the averaged image at a predetermined reduction ratio using a method, and an image for generating an enlarged image obtained by enlarging the reduced image at a predetermined enlargement ratio using a bilinear interpolation method An enlargement unit, a luminance distribution calculation unit that creates a luminance histogram of the enlarged image, a standard deviation calculation unit that analyzes the luminance histogram and calculates a standard deviation, and compares the standard deviation with a preset threshold value. Comparison judging means for judging the presence or absence of surface defects.

  According to a second aspect of the present invention, a light source unit that irradiates light onto the inspection surface, a light receiving unit that collects and reflects the light reflected from the inspection surface, and creates detection image data by imaging the imaged light A surface defect detection apparatus comprising: an imaging unit configured to perform image processing on the detected image data to determine whether there is a surface defect on the inspection surface, wherein the image processing unit includes the detected image data Using an averaging means for creating an averaged image obtained by arithmetic averaging, an image reducing means for creating a reduced image obtained by reducing the averaged image at a predetermined reduction ratio using a bilinear interpolation method, and a bilinear interpolation method An image enlarging unit that creates an enlarged image obtained by enlarging the reduced image at a predetermined magnification, a luminance distribution calculating unit that creates a luminance histogram of the enlarged image, and a standard that analyzes the luminance histogram and calculates a standard deviation Deviation calculation means And a comparison determination unit determines the presence or absence of surface defects by comparing the standard deviation with a preset threshold, are those provided.

  According to a third aspect of the present invention, an inspection surface is provided on a surface to be inspected using a surface defect detection device comprising an averaging means, an image reduction means, an image enlargement means, a luminance distribution calculation means, a standard deviation calculation means, and a determination means. A surface defect detection method for determining the presence or absence of a surface defect on the inspection surface based on detected image data obtained by imaging light reflected by the inspection surface by irradiating light, wherein the addition averaging means Creating an averaged image obtained by averaging the detected image data; and creating a reduced image obtained by reducing the averaged image at a predetermined reduction ratio using a bilinear interpolation method in the image reduction unit; A step of creating an enlarged image obtained by enlarging the reduced image at a predetermined magnification using a bilinear interpolation method, and a luminance histogram of the enlarged image created by the luminance distribution calculating unit. Do The step of analyzing the luminance histogram by the standard deviation calculating means to calculate the standard deviation, and the determination means comparing the standard deviation with a preset threshold value to determine whether there is a surface defect. And a step of determining.

  According to another aspect of the present invention, an averaged image is created by averaging the detected image data in a computer that has acquired the detected image data obtained by irradiating the inspection surface with light and imaging the light reflected by the inspection surface. Processing, processing for creating a reduced image obtained by reducing the averaged image at a predetermined reduction ratio using a bilinear interpolation method, and an enlarged image obtained by enlarging the reduced image at a predetermined enlargement rate using a bilinear interpolation method A process of creating a brightness histogram of the enlarged image, a process of calculating the standard deviation by analyzing the brightness histogram, and comparing the standard deviation with a preset threshold value to determine the surface defect. The process for determining the presence or absence is executed.

  As effects of the present invention, the following effects can be obtained.

  According to the present invention, noise caused by reflection from a metallic material or the like is removed from detected image data obtained by imaging reflected light from an inspection surface coated with a metallic material, and only a luminance change based on a small scratch group is obtained. Can be extracted. Therefore, it is possible to accurately detect surface defects due to a group of minute flaws, and it is possible to realize quantitative and objective automatic detection of surface defects without relying on the sensory evaluation of the operator.

Next, embodiments of the invention will be described.
FIG. 1 is a diagram showing the overall configuration of a surface defect detection apparatus according to an embodiment of the present invention, FIG. 2 is a block diagram showing the configuration of an image processing unit, and FIG. 3 is a flowchart of image processing. FIG. 4 is a diagram showing a process of image processing of detected image data on an inspection surface having no surface defect, FIG. 5 is a diagram showing a process of image processing of detected image data on an inspection surface having a surface defect, and FIG. It is a comparison figure of a standard deviation and a sensory evaluation result.
FIG. 7 is a comparison diagram of the standard deviation of the luminance distribution obtained by performing the conventional image processing and the sensory evaluation result.

[Surface defect detection device 11]
First, the configuration of the surface defect detection apparatus 11 according to an embodiment of the present invention will be described.
As shown in FIG. 1, the surface defect detection device 11 includes a light source unit 12, a light receiving unit 13, an imaging unit 14, and an image processing unit 15 that performs image processing on light detected by the imaging unit 14. .

The light source unit 12 is for irradiating the coating surface 19 as an inspection surface with light, and includes a light source. The light source is composed of, for example, an LED element, and light is applied from the light source to the painted surface 19 that is an inspection surface.
The light receiving unit 13 is for collecting the light emitted from the light source 21 and reflected by the painted surface 19, and is provided with a lens disposed above the painted surface 19 as an inspection surface. In addition to the lens, the light receiving unit 13 can also include a mirror for changing the direction of light.
The image pickup unit 14 picks up the light collected and imaged by the light receiving unit 13 and creates detection image data, and is a camera such as a CCD camera, for example.
The image processing unit 15 is for image processing the detected image data obtained by the imaging unit 14. In this image processing, the presence / absence of a surface defect such as a group of minute scratches on the painted surface 19 is determined.

  The image processing unit 15 is a single computer or a plurality of computers (electronic computers) in this embodiment, and includes an averaging means 41, an image reduction means 42, an image enlargement means 43, a luminance distribution calculation means 44, and a standard deviation calculation means 45. , And the determination means 46. However, the image processing unit 15 can also be configured by including dedicated devices that function as the respective units 41 to 46, respectively.

[Addition averaging means 41]
The averaging means 41, the detection image data obtained by imaging the inspection surface in the imaging unit 14 is for creating averaged images V ₁ integrated are averaged into one image data . The averaging means 41, a plurality of image data are aligned, are averaged, averaged image V ₁ is created. By averaging the image data, the image is averaged and noise for each pixel is removed.

[Image reduction means 42]
For example, as shown in FIGS. 5A and 5B, the image reduction means 42 is used to create a reduced image V ₂ obtained by reducing the averaged image V ₁ at a predetermined reduction magnification α. is there. The image reduction means 42 mathematically calculates a reduced image V _{2 obtained} by interpolating and reducing the averaged image V ₁ with the reduction magnification α using a bilinear interpolation method (bilinear interpolation method). The bilinear interpolation method is an interpolation method in which the value of a certain point is obtained from the values of four surrounding grid points.

[Image enlargement means 43]
The image enlarging means 43, for example, as shown in FIG. 5 (b) and FIG. 5 (c), the is used to create enlarged image V ₃ obtained by enlarging the reduced image V ₂ at a predetermined magnification β . The image enlarging means 43, by using bilinear interpolation method (bilinear interpolation), the reduced image V ₂ enlarged image V ₃ obtained by interpolation and expansion by the magnification β of is mathematically calculated.

  The reduction magnification α and the enlargement magnification β can be arbitrarily determined. However, when the reduction magnification α and the enlargement magnification β are multiplied, they have a relationship of 1 (for example, when the reduction magnification α is 10%, the enlargement magnification β is 1000%). This is desirable for the purpose of providing a correlation with the above.

[Luminance distribution calculating means 44 / standard deviation calculating means 45]
It said luminance distribution calculation means 44 calculates a luminance value of the enlarged image V _3, is used to create the luminance histogram.
The standard deviation calculating means 45 is for calculating the standard deviation σ by analyzing the luminance histogram.

[Determining means 46]
The determination means 46 is for determining the presence or absence of surface defects on the inspection surface based on the calculated standard deviation σ. The determination means 46 compares the preset threshold value σ ₀ with the standard deviation σ, and if the standard deviation σ exceeds the threshold value σ _{0, it} is determined that there is a surface defect (commonly called “white blur”). If the standard deviation σ is less than or equal to the threshold σ ₀ , it is determined that there is no surface defect.
The threshold value σ ₀ can be obtained experimentally or theoretically, and is set in the image processing unit 15 in advance.

[Configuration of Image Processing Unit 15]
FIG. 2 is a diagram illustrating an example of a hardware configuration of a computer 60 that functions as the image processing unit 15 according to the present embodiment.
The computer 60 as the image processing unit 15 includes a CPU 61, a ROM 62, a RAM 63, a storage unit 64, an input / output unit 65, an operation input unit 66, a display unit 67, etc., which are connected to each other via a bus 68. Therefore, data can be exchanged with each other under the management of the CPU 61.

The CPU 61 is a central processing unit that controls operation of the entire computer 60.
The ROM 62 is a read-only storage unit.
The RAM 63 is used when the CPU 61 executes various application programs stored in the storage unit 64 (programs for realizing the functions of the units 41 to 46 shown in FIG. 1 and the processing of the flowchart shown in FIG. 3). It is used as a work memory and also as a main memory used as necessary as a temporary storage area for various data.
The storage unit 64 is a storage unit that stores the various application programs and data.
The input / output unit 65 is a communication control unit for performing data transmission / reception with an external device via, for example, an arbitrary network. In this embodiment, the detected image data is captured from the imaging unit 14 via the input / output unit 65.
The operation input unit 66 is, for example, a keyboard, a mouse or the like, and the user operates these to input information to the computer.
The display unit 67 is a display and displays, for example, an averaged image V ₁ , a reduced image V ₂ , an enlarged image V ₃ , a luminance histogram, a determination result, and the like, which are calculation results of the image processing unit 15. is there.

[Image processing procedure]
Next, an image processing procedure for detected image data will be described with reference to the flowchart shown in FIG.
The detected image data is image data obtained using the light source unit 12, the light receiving unit 13, and the imaging unit 14 of the surface defect detection device 11 and using the metallic coating surface as an inspection surface. A plurality of detected image data is created for the same inspection surface, and the detected image data is acquired by the image processing unit 15 and image processing is performed.

The image processing unit 15 which has obtained the detected image data, at the averaging means 41, by averaging a plurality of detected image data to create an averaged image V ₁ (S21).

Then, the image reduction unit 42 of the image processing unit 15, as shown in FIG. 4 (a) (b) or FIG. 5 (a) (b), the averaged image _{V 1,} reduced by a predetermined reduction magnification α and, to create a reduced image _{V 2} (S22).
The image enlarging means 43 of the image processing unit 15, as shown in FIG. 4 (b) (c) or FIG. 5 (b) (c), the reduced image _{V 2,} enlarged by a predetermined magnification beta, to create an enlarged image _{V 3} (S23).

Subsequently, the luminance distribution calculation means 44 of the image processing unit 15 calculates the luminance value of the enlarged image V _3, for example, as shown in the right side of FIG. 4 or FIG. 5, to create the luminance histogram (S24). Further, the standard deviation calculating means 45 of the image processing unit 15 calculates the standard deviation σ of the luminance frequency distribution for the luminance histogram (S25).

Then, the determination unit 46 of the image processing unit 15 compares the calculated standard deviation σ with a preset threshold value σ ₀ (S26).
When the standard deviation σ exceeds the threshold σ ₀ (YES in S26), the determination unit 46 determines that there is a surface defect (S28).
Further, when the standard deviation σ is equal to or smaller than the threshold σ ₀ (NO in S26), the determination unit 46 determines that there is no surface defect (S27).

Finally, the image processing unit 15 outputs the determination result to the display unit 67 (S29), and ends the process.
Note that the image processing unit 15 displays one or a combination or all of the averaged image V ₁ , the reduced image V ₂ , the enlarged image V ₃ , and the luminance histogram each time the image or histogram is created. The unit 67 can display the output.

In FIG. 4, an averaged image V ₁ (FIG. 4 (a)), a reduced image V ₂ (FIG. 4 (b)), and an enlarged image obtained by using a painted surface (reference surface) having no micro-flaw group as an inspection surface. V ₃ (FIG. 4 (c)) and its standard deviation are shown.
Further, in FIG. 5, an averaged image V ₁ (FIG. 5A), a reduced image V ₂ (FIG. 5B), and an enlarged image V ₃ obtained by using a painted surface on which a minute flaw group exists as an inspection surface. (FIG. 5C) and its standard deviation are shown.
The painted surface shown in FIG. 4 or 5 is a metallic painted surface of an automobile, and the reduction magnification α in the image processing is 10%, and the enlargement magnification β is 1000%.

4 and 5, the detected image data obtained by imaging the reflected light from the inspection surface is subjected to image processing according to the present invention (for the averaged image V ₁ using a bilinear interpolation method). Interpolation reduction processing is performed, followed by interpolation expansion processing using a bilinear interpolation method), so that noise caused by reflection from the metallic layer of the metallic layer included in the detected image data is reduced. It can be seen that it has been removed. As a result of removing the noise in this way, in FIG. 4, the luminance change based on the micro-flaw group is emphasized and appears prominently.

Further, on the painted surface of FIG. 4 where there are no micro-scratches, the standard deviation of the luminance histogram is averaged image V ₁ ; 10.92> reduced image V ₂ ; 7.11> enlarged image V ₃ ; 5.62. On the painted surface where the micro-scratch group of FIG. 5 is present, the standard deviation of the luminance histogram is averaged image V ₁ ; 13.65> reduced image V ₂ ; 11.50> enlarged image V ₃ ; 10.65. It has changed.
That is, on the painted surface where there are no micro-flaw groups, image processing is performed to remove noise, and as a result, the standard deviation of the luminance histogram converges to a low value. In addition, even if image processing is performed and noise is removed on a painted surface where minute flaws are present, the characteristics of the luminance distribution (distribution with spread at the bottom and variation) are not lost, and the painted surface does not have minute flaws. The difference is clearer.
Therefore, the standard deviation of the luminance histogram is used as a reference value for determining the presence or absence of surface defects, so that surface defects due to minute flaws can be detected with high accuracy, and the operator's sensory evaluation is relied upon. No quantitative and objective automatic surface defect detection can be realized.

FIG. 6 is a diagram showing a correlation between the standard deviation σ calculated by the standard deviation calculating unit 45 and the sensory evaluation of the visual inspection performed by a skilled worker in the past in the present embodiment. In this figure, for each of the evaluation reference surface, the inspection surface determined to have no surface failure (OK) by sensory evaluation, and the inspection surface determined to have surface failure (NG) by sensory evaluation, The standard deviation σ calculated based on the processing method is plotted on the vertical axis.
From FIG. 6, it can be seen that there is a close correlation between the value determined by the sensory evaluation of the visual inspection and the value of the standard deviation σ of the luminance distribution of the site, which is determined to have a surface defect (a small flaw group). Recognize. Therefore, it can be said that the evaluation of the presence or absence of surface defects in the present embodiment is close to or higher than the sensory evaluation.

The figure which shows the whole structure of the surface defect detection apparatus which concerns on one Example of this invention. The block diagram which shows the structure of an image process part. The flowchart of image processing. The figure which shows the process of the image process of the detection image data of the test | inspection surface without a surface defect. The figure which shows the process of the image processing of the detection image data of the test | inspection surface with a surface defect. The comparison figure of the standard deviation of brightness distribution, and a sensory evaluation result. The comparison figure of the standard deviation of the luminance distribution obtained by performing the conventional image processing, and the sensory evaluation result.

Explanation of symbols

DESCRIPTION OF SYMBOLS 11 Surface defect detection apparatus 12 Light source part 13 Light receiving part 14 Imaging part 15 Image processing apparatus 41 Addition averaging means 42 Image reduction means 43 Image enlargement means 44 Luminance distribution calculation means 45 Standard deviation calculation means 46 Determination means

Claims (4)

An averaging means for creating an averaged image obtained by averaging the detected image data obtained by irradiating the inspection surface with light and imaging the light reflected by the inspection surface;
Image reduction means for creating a reduced image obtained by reducing the averaged image at a predetermined reduction magnification using a bilinear interpolation method;
An image enlarging means for creating an enlarged image obtained by enlarging the reduced image at a predetermined magnification using a bilinear interpolation method;
A luminance distribution calculating means for creating a luminance histogram of the enlarged image;
A standard deviation calculating means for analyzing the luminance histogram and calculating a standard deviation;
Comparison determination means for comparing the standard deviation and a preset threshold value to determine the presence or absence of surface defects,
A surface defect detection apparatus comprising: A light source unit that irradiates light onto the inspection surface; a light receiving unit that collects and focuses the light reflected from the inspection surface; an imaging unit that captures the imaged light to create detection image data; and A surface defect detection apparatus configured with an image processing unit that performs image processing on detected image data to determine the presence or absence of a surface defect on an inspection surface,
The image processing unit
An averaging means for creating an averaged image obtained by averaging the detected image data;
Image reduction means for creating a reduced image obtained by reducing the averaged image at a predetermined reduction magnification using a bilinear interpolation method;
An image enlarging means for creating an enlarged image obtained by enlarging the reduced image at a predetermined magnification using a bilinear interpolation method;
A luminance distribution calculating means for creating a luminance histogram of the enlarged image;
A standard deviation calculating means for analyzing the luminance histogram and calculating a standard deviation;
Comparison determination means for comparing the standard deviation and a preset threshold value to determine the presence or absence of surface defects,
A surface defect detection apparatus comprising: Using a surface defect detection apparatus comprising an averaging means, an image reduction means, an image enlargement means, a luminance distribution calculation means, a standard deviation calculation means, and a determination means, the inspection surface is irradiated with light and the A surface defect detection method for determining the presence or absence of a surface defect on the inspection surface based on detected image data obtained by imaging light reflected by the inspection surface,
Creating an averaged image obtained by averaging the detected image data by the averaging means;
Creating a reduced image obtained by reducing the averaged image at a predetermined reduction ratio using a bilinear interpolation method in the image reduction means;
Creating an enlarged image obtained by enlarging the reduced image at a predetermined enlargement ratio using a bilinear interpolation method in the image enlarging means;
Creating a brightness histogram of the enlarged image in the brightness distribution calculating means;
Analyzing the luminance histogram and calculating a standard deviation in the standard deviation calculating means;
In the determining means, comparing the standard deviation with a preset threshold value to determine the presence or absence of surface defects,
A surface defect detection method comprising: To the computer that acquired the detected image data obtained by irradiating the inspection surface with light and imaging the light reflected by the inspection surface,
A process of creating an averaged image obtained by averaging the detected image data;
Processing to create a reduced image obtained by reducing the averaged image at a predetermined reduction ratio using a bilinear interpolation method;
A process of creating an enlarged image obtained by enlarging the reduced image at a predetermined enlargement magnification using a bilinear interpolation method;
A process of creating a brightness histogram of the enlarged image;
Processing for analyzing the luminance histogram and calculating a standard deviation;
A process of comparing the standard deviation with a preset threshold value to determine the presence or absence of surface defects,
A surface defect detection program characterized by being executed.