JP2009204473A - Painted state inspecting device and inspection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a painted state inspecting device and an inspection method, capable of stably performing high-reliability smoothness inspection by a detection value for precisely reflecting the state of a painted surface, without going through complex image processings. <P>SOLUTION: This device is equipped with an image projection means 2 for projecting onto a surface 40 to be inspected on the painted surface 4; a bright and dark image 20 constituted of an overall dark part 20b and thin line-shaped bright parts 20a arranged in the dark part; an imaging means 3 for imaging a projection image 50 of the surface 40 to be inspected; an image processing means 5 for binarizing an acquired multi-gradation image 50, with a prescribed threshold V<SB>T</SB>, and extracting a brightness/darkness boundary line 52c in the binary image; and a determination means for determining smoothness of the painted surface, by analyzing a waveform of the brightness/darkness boundary line. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an apparatus and an inspection method for inspecting a finished state of a painted surface.

  Conventionally, as a method for inspecting the finished state of a painted surface of an automobile or a railway vehicle, for example, a black and white vertical stripe grid is projected onto the painted surface, and the smoothness of the painted surface is determined from the disturbance of the width of the bright part in the reflected image. There have been proposed a method (Patent Document 1) and a method (Patent Documents 2 and 3) for determining the smoothness of a painted surface from disturbances in the width and area of a boundary region (intermediate gradation) between a bright part and a dark part.

  However, since the bright and dark patterns of the black and white vertical stripes employed in these inspection methods both have a high proportion of bright portions, the overall luminance increases due to the diffuse reflection component in the bright portions, which is sufficient for the bright and dark portions. There is a problem that contrast cannot be obtained (see FIG. 3B). In particular, this tendency becomes remarkable on a light-colored painted surface such as white, and there is a possibility that the quality of the painted state may not be reflected in the inspection result.

  In addition, the method of Patent Document 1 has a fundamental problem that the portion where both sides of the bright portion fluctuate in the same phase cannot be distinguished from the case where the width does not fluctuate. Further, in the methods of Patent Documents 2 and 3, a portion having a large luminance gradient that shifts from a bright part to a dark part is extracted as a boundary region. Since this is detected, the same problem as in the above-mentioned Patent Document 1 results. In Patent Document 3, an exclusive OR is performed with the data from which noise has been removed by the expansion / contraction process. There is no denying that it will affect the test results.

Japanese Patent Publication No. 6-19262 Japanese Patent No. 2508176 Japanese Patent No. 3412366

  The present invention has been made in view of such a situation of the prior art, and the object thereof is to perform smooth smoothness with high reliability by a detection value accurately reflecting the state of a painted surface without going through complicated image processing. The present invention provides a coating state inspection apparatus and an inspection method capable of stably performing a property inspection.

In order to solve the above-described problems of the prior art, the present invention provides:
A coating state inspection device for inspecting the finished state of a painted surface,
Image projection means for projecting a light / dark image composed of a whole dark part and a thin line-shaped bright part arranged in the dark part on the surface to be inspected on the painted surface;
Imaging means for imaging a projected image of the surface to be inspected;
Image processing means for binarizing a multi-tone image obtained by the imaging means with a predetermined threshold value, and extracting a light-dark boundary line in the binary image;
Determination means for analyzing the waveform of the light-dark boundary line to determine the smoothness of the painted surface;
It is in the coating state inspection apparatus provided with.

  In the present invention, it is preferable that a plurality of the thin line-like bright portions of the bright and dark image are arranged at an equal pitch in the dark portion. Furthermore, it is preferable that the width of the thin linear light portion is 0.7 mm or less and 1/16 or less of the pitch.

The present invention also provides:
A coating state inspection method for inspecting the finished state of a painted surface,
Projecting on the surface to be inspected on the painted surface a light / dark image composed of a full dark part and a thin bright line arranged in the dark part,
The projected image of the surface to be inspected is imaged by an imaging means,
Binarizing a multi-tone image obtained by the imaging means with a predetermined threshold, and extracting a light and dark boundary line in the binary image;
It is in the coating state inspection method including analyzing the waveform curve of the light / dark boundary line and determining the smoothness of the painted surface.

  In the present invention, it is preferable that the analysis of the waveform curve includes obtaining a maximum amplitude in a predetermined section of the light / dark boundary line and comparing it with a reference value. Further, the analysis of the waveform curve obtains the center line of the light / dark boundary line, divides the area surrounded by the center line and the light / dark boundary line by the section length, obtains the average amplitude, and compares it with the reference value It may include.

  Further, in the present invention, in the binarization process, the pixels of the multi-tone image are sorted based on the luminance, and the luminance at the position of 1/3 from the lowest luminance is assumed to be the dark portion average luminance. A first standard value obtained by calculating a standard deviation of luminance below that and adding a value obtained by multiplying the standard deviation by 3 to the dark part average luminance, and a second value obtained by adding a predetermined offset value to the dark part average luminance; Preferably using the larger of them as the threshold value.

  In the binarization process, the multi-gradation image is linearly converted and standardized so that the luminance distribution of the multi-gradation image is distributed between the maximum luminance and the minimum luminance in the image format. It may include using an intermediate luminance value as the threshold value.

  Since the coating state inspection apparatus and the inspection method according to the present invention are configured as described above, they have the effects described below.

  First, by using a bright / dark image composed of an entire dark portion and a thin line-shaped bright portion arranged in the dark portion as the inspection pattern, the proportion of the bright portion in the bright / dark image is reduced, and the projected image The increase in overall brightness is suppressed, and high contrast is obtained in the bright area with respect to the entire dark area, resulting in disturbance of diffused light appearing on both sides of the thin line-shaped bright area, that is, due to the surface properties of the painted surface The irregular reflection thus made can be detected with high accuracy (see FIG. 3A). As a result, even on a light-colored painted surface such as white, an increase in the brightness of the entire image due to the painted color is suppressed, a high contrast is obtained between the bright and dark portions, and high-precision inspection is possible.

  Second, a multi-tone image obtained by capturing a projected image is binarized with a predetermined threshold value, and a light and dark boundary line in the binary image is extracted, so that the painted surface is not subjected to complicated image processing. It is possible to easily and accurately detect only the irregular reflection of the image, and it is possible to avoid the factor resulting from the image processing algorithm from being mixed into the inspection result.

  Third, irregular reflection caused by the surface properties of the painted surface appears as the amplitude and frequency of the light / dark boundary line (waveform curve) and their fluctuations, so the smooth and linear light / dark boundary with low amplitude and frequency. A good painted surface with a line can be easily and reliably detected on a poorly painted surface with large amplitude and high frequency, and the maximum amplitude value per unit length, frequency, or the center line of the waveform curve The smoothness of the painted surface or the degree of painting failure can be determined by various methods such as the areas of the portions appearing on both sides.

  Fourthly, by separately extracting the light and dark boundary lines appearing on both sides of the thin line-shaped bright portion, even the portion where they vibrate in the same phase can be reliably detected as the amplitude and the vibration direction. In the first place, irregular reflection appears as an increase in luminance in the dark part, but further increase in luminance in the bright part is difficult to detect. On the other hand, in the present invention, it is possible to detect a decrease in luminance in the bright portion by suppressing the luminance of the entire projected image by the thin line-shaped bright portion.

Hereinafter, the present invention will be described in detail based on illustrated embodiments.
FIG. 1 is a side sectional view showing an outline of a coating state inspection apparatus according to an embodiment of the present invention. In the figure, an inspection apparatus 1 includes an image projecting unit 2 that projects a predetermined light and dark pattern onto an inspected surface 40 of a painting surface 4, an imaging unit 3 that captures a projected image projected onto the inspected surface 40, and an imaging unit. 3 is mainly composed of an image processing means 5 which also serves as a determination means for processing the image obtained by 3 to determine smoothness.

  The image projection means 2 is configured by arranging a light source 21 (backlight) behind a screen 20 (bright / dark pattern) via a polarizing plate 23 and a diffusion plate 22 and covering them with a case. The illumination light is diffused by the diffusing plate 22 so that the luminance distribution of the entire screen is made uniform. Further, only the linear component of the illumination light is transmitted through the screen 20 by the polarizing plate 23, and the light / dark pattern of the screen 20 is projected onto the inspection surface 40 Is possible.

  The imaging means 3 is preferably a CCD camera that can capture a projected image as two-dimensional array of digital image data. This image pickup means 3 is arranged in the case 11 (hood) together with the image projection means 2, and if the bottom of the case 11 provided with the opening 10 is positioned in contact or non-contact with the painted surface 4, the image is obtained. The screen 20 (light / dark pattern) of the projection unit 2 is projected onto the surface 40 to be inspected facing the opening 10, and the regular reflection can be captured by the imaging unit 3.

  The image processing means 5 includes a central processing unit (CPU), a storage device, a screen display device (monitor), an input device (keyboard, pointing device, etc.), etc., not shown, and can execute each image processing program described later. Yes, it can be configured by a dedicated microcomputer, or a PC (Personal Computer) operating on a general-purpose operating system can be used. These may also serve as control means for controlling the image projection means 2 and the imaging means 3.

  FIG. 2 shows a light / dark pattern of the screen 20. In the figure, the screen 20 is formed by arranging a plurality of thin line-shaped bright portions 20a in parallel at an equal pitch on the entire dark portion 20b, for example, on a polarizing plate 23 or a glass plate covering the surface thereof. It can be formed by adhering, sticking, or applying a belt-like coating constituting each dark portion 20b, or by providing a slit corresponding to each bright portion 20a in the light-shielding sheet material.

  In order to keep the brightness of the entire projected image on the screen 20 low, it is necessary to make the width W of the bright portions 20a as small as possible and to make the pitch P of the bright portions 20a sufficiently large with respect to the width W. 3 (a) and 3 (b) are diagrams schematically showing images 50 and 50 'obtained by light and dark patterns having different bright portion widths and pitches, and luminance distributions thereof, respectively. In the conventional light / dark pattern 50 'in which the widths of the bright part 50a' and the dark part are substantially equal to each other, the brightness of the entire image increases due to diffuse reflection and irregular reflection in the bright part 50a ', and the bright part 50a' The contrast C ′ becomes smaller.

  On the other hand, in the light / dark pattern 50 of the present invention in which the width of the bright portion 50a is narrow as shown in FIG. 3A, the luminance of the entire image can be kept low, so that a high contrast C between the bright portion 50a and the dark portion. Thus, it is possible to reliably reflect the disturbance of diffuse reflection near the boundary of the bright part 50a in the binary image described later.

  For the reasons described above, the width W of the bright portion 20a is preferably 1 mm or less, more preferably 0, although it depends on the color tone and processing area of the coating surface 4 to be inspected, the distance from the screen 20, and the like. 0.7 mm or less, and preferably 1/16 or less of the pitch P. In order to efficiently arrange the bright portion 20a serving as the detection region in the limited screen 20, the width of the dark portion may be less than the pitch P in the peripheral portion of the screen 20. Even in that case, it is preferable that the total area of the bright portion 20a is 7% or less of the entire screen.

  Next, a method for inspecting the smoothness of a painted surface based on the above embodiment will be described with reference to the drawings.

  The light / dark pattern of the screen 20 projected onto the surface to be inspected 40 on the painted surface 4 by the image projecting means 2 is imaged as a multi-gradation image 50 (in the embodiment, a grayscale image of 256 gradations) by the imaging means 3. . FIG. 4A is an enlarged view of a poorly painted portion detected in the actually captured image 50. In the drawing, for convenience, a binary image is obtained by dithering, but it can be confirmed that the overall luminance is kept low and that the outline of the thin line-shaped bright portion (50a) is disturbed although the luminance is low.

When the multi-tone image 50 is binarized with a predetermined threshold (V _T ) described later, a binary image 51 shown in FIG. 4B is obtained. By performing edge detection processing on the binary image 51, only the light and dark boundary lines of the binary image 51 can be extracted as the waveform curve 52 (52c). The relationship between the waveform curve 52 and the minute unevenness of the painted surface 4 can be considered as follows.

  In the inspection apparatus 1 of FIG. 1, incident light incident obliquely on the surface 40 to be inspected is specularly reflected at the same reflection angle as the incident angle, but the painted surface 4 is not a perfect mirror surface and is a diffusive object. Therefore, the reflected light contains a diffusing component. However, since the diffuse reflection also occurs to some extent if it is a good painted surface, the waveform curve 52 is a curve with little disturbance and therefore a small amplitude.

  Here, for easy understanding, when viewed from a direction orthogonal to FIG. 1, the thin-line bright portion 20a (part) of the screen 20 shown in FIG. When projected, it becomes irregularly reflective depending on the degree of unevenness. That is, the expansion of the bright portion 20a in the both width directions due to the diffuse reflection varies according to the irregular reflection, and the degree of the variation is an index reflecting the quality of the painted surface 4.

Next, processing for determining a threshold value (V _T ) for binarizing the multi-tone image 50 will be described. The threshold value (V _T ) for binarizing the multi-tone image 50 varies depending on the color tone of the painted surface 4 and the like. Therefore, an optimum threshold value (V _T ) is determined from the multi-tone image 50 by the following processing.

First, the pixels of the multi-tone image 50 are sorted based on the luminance, and the luminance (V _B ) at a position 1/3 from the lowest luminance is assumed to be the average luminance in the dark portion, and is equal to or lower than the luminance (V _B ). The luminance standard deviation (V _D ) is obtained. Next, a value obtained by multiplying the standard deviation σ by 3 and the luminance (V _B ) at the 1/3 position is defined as a first threshold value (V _B + 3 · V _D ).

The first threshold value can be said to be a value obtained by setting the boundary between the light and dark areas from the low luminance side in consideration of the variation in the luminance up to the value assumed to be the average luminance (V _B ) of the dark area (51b). However, when the luminance variation is extremely small, the first threshold value (V _B + 3 · V _D ) may deviate from the light / dark boundary. Therefore, a value obtained by adding a predetermined offset value (V _S ) to the luminance value (V _B ) at the 1/3 position is set as a second threshold (V _B + V _S ), and the first threshold (V _B +3) is set. The larger of V _D ) and the second threshold (V _B + V _S ) is used as the threshold. Therefore, the threshold value (V _T ) is set by the following equation.
V _T = max (V _B + 3 · V _D , V _B + V _S )

FIG. 5C shows the luminance distribution in the vicinity of the bright portion 50a of the multi-tone image 50. With the threshold value (V _T ) set as described above, the binary image 51 has a luminance due to the influence of irregular reflection. It reflects the light and dark border that turns up. FIG. 5D shows a binary image 51 of the same portion binarized by the threshold value (V _T ).

  Further, FIG. 5 (e) shows the light / dark boundary line 52 of the same part extracted by the edge detection processing, and the light / dark boundary line 52 has two widths extending along both sides of the light portion 51a. There is no waveform curve 52c (having a width corresponding to one pixel), and the section where the disturbance of each waveform curve 52c is large is greatly affected by irregular reflection in the bright part, and there is a problem in the smoothness of the painted surface 4 It can be said that it is a part. Therefore, the smoothness of the painted surface 4 (surface to be inspected 40) can be determined by detecting a section where the amplitude of the waveform curve 52c is large or a section where the frequency of the relatively large amplitude is large.

  For example, as shown in FIG. 5 (e), the maximum amplitude R in a predetermined section in the extending direction of the waveform curve 52c is obtained and compared with a reference value. It can be determined that there is. In actual processing, a plurality of fine line-shaped bright portions 20a are set for one screen 20, and the waveform curve 52c on both sides of each bright portion 20a exceeds a predetermined section length in the vertical axis direction in the figure. The maximum amplitude R in each section is calculated by sequentially applying while wrapping, and compared with a reference value.

  Alternatively, the center line of the waveform curve 52c may be obtained, and the area surrounded by the center line and the waveform curve 52c may be divided by the section length to obtain the average amplitude and compared with the reference value. In that case, the center line can be detected as a straight line, but it can also be detected as a curve by setting a lower limit value for the curvature. In the latter case, the surface to be inspected 40 may include a curved surface. Yes.

In the above-described embodiment, the case where the threshold value (V _T ) for binarizing the multi-tone image 50 is determined by a method that assumes the average luminance of the dark part is shown. However, the threshold value (V _T ) is set. Prior to this, the multi-tone image 50 may be linearly converted and standardized so that the luminance distribution is distributed between the maximum luminance and the minimum luminance in the image format. When the multi-tone image 50 is a grayscale image with 256 tones, the minimum luminance is 0 and the maximum luminance is 255. By performing such standardization processing, a fixed threshold value can be used, or a value obtained by adding calibration to the threshold value in consideration of the standard deviation as described above can be used as the threshold value.

As mentioned above, although embodiment of this invention was described, this invention is not limited to the said embodiment, A various deformation | transformation and change are possible based on the technical idea of this invention.
For example, in the above-described embodiment, the case where the image projecting unit 2 and the imaging unit 3 are disposed in the case 11 has been shown. Alternatively, the inspection room itself can be inspected while preventing reflection on the surface to be inspected 40. Further, a disturbance light can be completely blocked by providing a flexible light shielding member around the opening 10 of the case 11.

It is a sectional side view which shows the outline of the coating state inspection apparatus which concerns on this invention embodiment. It is a figure which shows the screen (brightness pattern) which concerns on this invention embodiment. (A) is a schematic diagram which shows the light-dark pattern and its luminance distribution based on this invention, (b) is a schematic diagram which shows the conventional light-dark pattern and its luminance distribution method. The flow of the image processing in the paint state inspection which concerns on this invention is shown, (a) is a multi-tone image, (b) is a binary image, (c) is a principal part enlarged view which shows a light-dark boundary line. It is a schematic diagram which shows each step of the image processing in the coating state test | inspection which concerns on this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Inspection apparatus 2 Image projection means 3 Imaging means 4 Paint surface 5 Image processing means 10 Opening 11 Case 20 Screen (brightness pattern)
20a, 50a, 51a bright portions 20b, 50b, 51b dark portion 40 to be inspected surface 50 multi-tone image 51 the binary image 52 dark boundary line 52c waveform curve _{V T} threshold _{V B} dark portion average brightness R maximum amplitude

Claims (8)

A coating state inspection device for inspecting the finished state of a painted surface,
Image projection means for projecting a light / dark image composed of a whole dark part and a thin line-shaped bright part arranged in the dark part on the surface to be inspected on the painted surface;
Imaging means for imaging a projected image of the surface to be inspected;
Image processing means for binarizing a multi-tone image obtained by the imaging means with a predetermined threshold value, and extracting a light-dark boundary line in the binary image;
Determination means for analyzing the waveform of the light-dark boundary line to determine the smoothness of the painted surface;
A coating state inspection device with   The coating state inspection apparatus according to claim 1, wherein a plurality of the thin line-shaped bright portions of the light-dark image are arranged at an equal pitch in the dark portion.   The coating state inspection apparatus according to claim 2, wherein a width of the thin linear light portion is 0.7 mm or less and 1/16 or less of a pitch. A coating state inspection method for inspecting the finished state of a painted surface,
Projecting on the surface to be inspected on the painted surface a light / dark image composed of a full dark part and a thin bright line arranged in the dark part,
The projected image of the surface to be inspected is imaged by an imaging means,
Binarizing a multi-tone image obtained by the imaging means with a predetermined threshold, and extracting a light and dark boundary line in the binary image;
A paint state inspection method, comprising: analyzing a waveform curve of the light / dark boundary line to determine smoothness of a paint surface.   The paint state inspection method according to claim 4, wherein the analysis of the waveform curve includes obtaining a maximum amplitude in a predetermined section of the light / dark boundary line and comparing it with a reference value.   The analysis of the waveform curve determines the center line of the light / dark boundary line, divides the area surrounded by the center line and the light / dark boundary line by a section length to obtain an average amplitude, and compares it with a reference value. The coating state inspection method according to claim 4 comprising.   In the binarization process, the pixels of the multi-tone image are sorted based on the luminance, and the luminance at a position 1/3 from the lowest luminance is assumed to be the dark portion average luminance, and the luminance standard lower than that is assumed. A deviation is obtained, and a first value obtained by multiplying the standard deviation by 3 is added to the dark part average brightness, and a second value obtained by adding a predetermined offset value to the dark part average brightness is obtained. The coating state inspection method according to any one of claims 4 to 6, comprising using a larger one as the threshold value.   In the binarization process, the multi-gradation image is linearly converted and standardized so that the luminance distribution of the multi-gradation image is distributed between the maximum luminance and the minimum luminance in the image format. The paint state inspection method according to any one of claims 4 to 6, comprising using an intermediate luminance value as the threshold value.