JPH0772909B2 - Welding condition judgment method by visual inspection - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a welding condition judging method by visual inspection for judging the quality of a welding condition by subjecting an image of an inspection object, which is a welded object, to image processing by a computer. .

[0002]

2. Description of the Related Art Conventionally, an image input device such as an ITV camera captures an image of a spatial region including an inspection object which is a welded product,
A method has been proposed in which the quality of a welding state is determined by performing image processing by a computer on an image input from an image input device (see Japanese Patent Laid-Open No. 181989/1989). This method is to inspect the welding state after laser welding, irradiating the strip-shaped parallel rays on the surface of the inspection object obliquely from above, and the portion where the strip-shaped parallel rays are irradiated is the surface of the inspection object. An image is captured from above in the vertical direction, and the image thus obtained is subjected to image processing to determine the quality of the welding state. That is, it is possible to capture the image of the unevenness of the surface of the inspection object from the image input device by irradiating the strip-shaped parallel rays obliquely above the surface of the inspection object, and You can get information.
Therefore, it is possible to detect defects such as sink marks and insufficient width of the welded portion based on the captured image.

[0003]

However, in the above method, since the linearity of the strip-shaped parallel rays and the uniformity of the illuminance affect the inspection accuracy, the optical system for irradiating the strip-shaped parallel rays is complicated. There is a problem of becoming a configuration. In addition, since the map including the height information is captured in the image input device, the strip parallel rays with respect to the surface of the inspection object and the angle of the optical axis of the image input device affect the inspection accuracy,
There is also a problem that the processing process becomes complicated because the image processing is performed on the image captured from the image input device by including such angle information.

The present invention is intended to solve the above-mentioned problems, prevents the light source for irradiating the surface of the inspection object from affecting the inspection accuracy, and performs processing relating to height information of the surface of the inspection object. It is intended to provide a welding state determination method by visual inspection that simplifies the processing process by performing image processing that does not include.

[0005]

According to a first aspect of the present invention, a space region including a welded portion of an inspection object, which is obtained by superposing a plurality of workpieces and welding them by laser welding, is imaged by an image input device, and then an image is obtained. In a welding state determination method by visual inspection that determines the quality of the welding state by performing processing, based on an original image that is a grayscale image input from an image input device, a differentiation corresponding to the rate of change in density with adjacent pixels A differential absolute value image in which the absolute value is the value of each pixel, a differential direction value image in which the differential direction value corresponding to the changing direction of the density with an adjacent pixel is the value of each pixel, and an edge indicating the boundary line of the density change After creating an image and setting an inspection area including a welding point, a pixel whose differential absolute value is larger than a predetermined threshold value for a pixel on a boundary line in an edge image in the inspection area is obtained as a welding feature pixel, A frequency distribution relating to the differential direction value of contact feature pixel, by comparing the frequency distribution of the non-defective are preset, it is to judge the quality of the welding conditions.

According to the second aspect of the present invention, the inspection area is set so as to include a welded portion and a non-welded portion on one of the workpieces, and a predetermined range of the differential direction value for the welding feature pixel. The welding state is determined to be good when the sum of the frequencies is higher than a threshold value set in advance as the sum of the frequencies of non-defective products in the same range. In the invention of claim 3, the inspection area is set inside the welding spot, and the sum of the frequencies of the differential direction value of the welding feature pixel in the predetermined range is greater than the preset threshold value as the sum of the frequencies of non-defective products in the same range. Is also large, it is determined that the welding state is poor.

[0007]

According to the method of claim 1, since the frequency distribution regarding the differential direction value is obtained for the pixels on the boundary line of the unevenness formed by laser welding and compared with the frequency distribution of the non-defective product, the positional relationship between the light source and the image input device is obtained. Does not need to have a positional relationship that includes height information of the surface of the inspection object, and may be set so that a shadow corresponding to the unevenness of the welding location is formed. That is, since no special light source for irradiating the band-shaped parallel rays is required, the structure of the light source is simplified and the restrictions on the arrangement of the light sources are reduced. Also, with regard to image processing, there is no need to perform complicated calculations that include information about the height of the surface of the inspection target. In addition to general processing such as differential processing and thinning processing, addition and simple comparison of size comparison Therefore, the processing steps of the image processing are simplified.

The methods of claims 2 and 3 are desirable embodiments, and the quality of the inspection can be easily determined by devising the method of setting the inspection area, and the differential direction value of the inspection object can be easily determined. Since the frequency distribution and the frequency distribution of non-defective products are compared by a simple process of limiting the range of differential direction values and comparing the frequency of that range with a threshold value, it is possible to judge the quality of the welded part. The process of the image processing for performing the process becomes simpler.

[0009]

【Example】

(Embodiment 1) In this embodiment, a case of detecting a defective welding state in laser spot welding will be described. As shown in FIG. 2, a pair of workpieces 11a and 11b are overlapped with each other, and a laser beam is applied so as to straddle the contact portions of the workpieces 11a and 11b, so that the surface is exposed as shown in FIG. Both workpieces 1 with the recess 12 formed
1a and 11b are welded (see FIG. 4). The surface condition of the workpieces 11a and 11b, which are inspection objects, is captured by the image input device 1 such as an ITV camera, and the light source reflected by the surfaces of the workpieces 11a and 11b before welding (see FIG. The light sources are arranged in such a positional relationship that specularly reflected light (light rays are shown by arrows) from (not shown) enter the image input apparatus 1. Therefore, after welding, regular reflection light cannot be obtained in the recess 12, and the recess 12 formed by welding and the other portion can obtain contrast in the image captured from the image input device 1. In this embodiment, the quality of the welded state is determined by using the contrast of the image formed in this way.

As shown in FIG. 9, the density of each pixel of the image captured by the image input device 1 is converted into a digital signal by the analog-digital converter 2, and the following pre-processing is performed by the pre-processor 3. As a result, in addition to the original image obtained as the output of the analog-digital conversion unit 2, a differential absolute value image, a differential direction value image, and an edge image are obtained.

That is, the original image obtained by imaging the spatial region including the inspection object is a grayscale image, and the processing for extracting the edge such as the contour line of the inspection target from the grayscale image is
It is based on the idea that "the edge corresponds to a portion where the density change is large". Therefore, the edge is extracted by differentiating the density. The differential processing is shown in Fig. 1.
As shown in 0, the original image P ₁ is divided into 3 × 3 pixel local parallel windows W. That is, with the pixel E of interest in the original image P ₁ as the center, 8 pixels surrounding the pixel E
The pixels (near 8) A to D and F to I form a local parallel window W, and the vertical density change ΔV and the horizontal density change ΔH of the density of the pixels A to I in the local parallel window W are defined. Calculate by the following formula.

ΔV = (A + B + C)-(G + H + I) ΔH = (A + D + G)-(C + F + I) where A to I represent the densities of the corresponding pixels. Further, the differential absolute value abs (E) and the differential direction value deg (E) for the pixel E are obtained by the following equation. abs (E) = (ΔV ² + ΔH ² ) ^1/2 deg (E) = tan ⁻¹ (ΔV / ΔH) + (π / 2) The differential absolute value abs (E) is the pixel E of interest in the original image. Represents the rate of change of density in the vicinity area, and the differential direction value
Indicates a direction orthogonal to the direction of density change in the same neighborhood area. By performing the above calculation for all the pixels of the original image P ₁ , it is possible to extract a portion such as the contour line of the inspection object where the density change is large and the direction of the change. Here, the value of each pixel is the differential absolute value
Image that is abs (E) is the differential absolute value image, differential direction value deg
The image of (E) is called a differential direction value image.

Next, thinning processing is performed. The thinning process is performed by paying attention to the fact that the larger the differential absolute value, the larger the density change. That is, the differential absolute value of each pixel is compared with the differential absolute value of the surrounding pixels, and those having a larger value than the surrounding pixels are connected to extract an edge having a width of one pixel. As shown in FIG. 11, considering the differential absolute value image P _{2 in} which the position of each pixel is represented by XY coordinates and the differential absolute value is represented by Z coordinates, the thinning process is to find the ridgeline on this curved surface. Equivalent to. By the processing up to this point, all ridgelines are extracted regardless of the magnitude of the differential absolute value. Since the ridge line obtained at this stage also includes unnecessary small peaks due to noise or the like, as shown in FIG. 12, a threshold SL is appropriately set, and only a value equal to or higher than this threshold SL is adopted. Remove noise components. The image obtained by this process is
When the contrast of the original image is insufficient or when there is much noise, discontinuous lines are likely to occur. Therefore, edge extension processing is performed. In the edge extension processing, starting from the end point of the discontinuous line, the pixel of interest and the surrounding pixels are compared, the line is extended in the direction in which the evaluation function f (J) represented by the following equation becomes the largest, and the other Continue doing this until you hit the end of the line.

F (J) = abs (J) · cos (deg (J) −deg (0)) · cos ((J−1) π / 4−deg (O)) where deg (O) is Center pixel (E of the local parallel window W
Deg (J) is the differential direction value of an adjacent pixel (corresponding to 8 neighborhoods of the local parallel window W), ab
s (J) is the differential absolute value of the adjacent pixel. However, J = 1,
2, ..., 8.

By the above processing, an edge image that traces the contour of the portion of the original image P _{1 where} the density change is large can be obtained. Thus, as shown in FIG. 13, four types of images of the original image P ₁ , the differential absolute value image P ₂ , the differential direction value image P ₃ , and the edge image P ₄ are obtained, and each image is acquired by the original image memory 4, the differential image. It is stored in the absolute value image memory 5, the differential direction value image memory 6, and the edge image memory 7, respectively. In the following description, the pixel positions of the images P _{1 to} P ₄ are represented by XY coordinates, and the pixel values in each image are f ₁ (x, y) and f ₂ (x, y), respectively. , F ₃ (x, y), f
₄ (x, y).

The original image is a grayscale image, and the density a (= f ₁ (x, y)) of each pixel is represented by, for example, 8 bits and 0 ≦ a ≦ 255. Further, the differential absolute value b (= f ₂ (x, y)) of each pixel of the differential absolute value image is represented by 6 bits, for example, 0 ≦ b ≦ 63, and the differential direction of each pixel of the differential direction value image The value c (= f ₃ (x, y)) is, for example, 1
Expressed in six directions, 0 ≦ c ≦ 15. As for the edge image, only the presence / absence of a line is present, and therefore the value of the pixel forming the line is “1” and the value of the other pixels is “0”. That is,
The range of f ₄ (x, y) is {0, 1}. In the following description, the term density represents white density, and the larger the density value, the brighter.

The arithmetic processing section 8 is a section for performing image processing by a microcomputer, and performs the processing shown in FIG. That is, as shown in FIG. 5, based on the original image stored in the original image memory 4, one of the workpieces 11 a, which is set so that the specular reflection light enters the image input device 1 before welding, is displayed. The inspection area M ₁ is set in the above-mentioned bright and dark portion (step S1). The inspection region M ₁ can be set in any shape as long as it is set so as to straddle the welded portion and the non-welded portion on the one workpiece 11a. Next, as shown by the arrow in FIG. 6, raster scanning is performed on the data stored in the edge image memory 7 in the set inspection area M ₁ (step S2), and the value is “1”.
Is detected as an edge flag point. With respect to the edge flag point thus obtained, the value (differential absolute value) of the pixel having the same coordinates is obtained from the differential absolute value image memory 5, and the obtained differential absolute value A and the predetermined threshold value SL1 are compared in size (step). S4). Here, when the condition of A> SL1 is satisfied, it is estimated that the point is on the boundary line of the recess 12 because it is a portion where the contrast is large. The pixel thus obtained will be referred to as a welding feature pixel. Therefore, the value of the welding feature pixel (differential direction value) is read from the differential direction value image memory 6 and the frequency for each differential direction value is integrated (step S5). That is, when a welding feature pixel is found, it is classified for each differential direction value and the number is incremented. The above process is repeated until all the pixels in the inspection area M ₁ have been scanned (step S
6) Obtain the frequency distribution regarding the differential direction value of the welding feature pixel.

After the frequency distribution is obtained in this way, the quality of the welded state is determined by comparing it with the frequency distribution of a non-defective product registered in advance. When welding is performed, the contrast is increased due to the formation of the recess 12, so that the number of welding feature points increases and the number of welding feature points increases, as shown in FIG.
The frequency of the differential direction value increases as shown in. In the case of non-welding or welding failure, since the recess 12 is not formed, the contrast of the original image is small and the number of welding characteristic pixels is small. Therefore, as shown in FIG. 7B, the frequency of the differential direction value becomes small. Further, when welding is performed, the frequencies of the differential direction values are concentrated in two poles. If the welding state is good, the concentration is low, and if the welding is poor, the concentration is high. Based on such knowledge, a range is set for two poles in which the frequencies are concentrated with respect to the differential direction value, and the sums α and β of the frequencies in each range are compared with predetermined thresholds SL2 and SL3 (step S7, S8). Here, when α> SL2 β> SL3 are both satisfied, it is determined that the welding state is good, and when either one is not satisfied, it is determined that the welding condition is bad.

(Embodiment 2) In Embodiment 1, since the inspection area M ₁ is set on one of the workpieces 11a, even if the welding state is defective, it may not be recognized as a good product. is not. Therefore, in this embodiment, as shown in FIG. 8, while straddling over both the workpieces 11a and 11b,
An inspection area M ₂ is set in the recess 12. It is known that the depth of the welded portion is about 100 μm when the welding state is good, but it is about 20 to 30 μm when the welding state is defective. In this case, if the welding state is defective, the specularly reflected light from the vicinity of the bottom of the recess 12 increases and the contrast with the vicinity of the periphery of the recess 12 becomes stronger. In this embodiment, this is used to judge the quality of the welding state. That is, after obtaining the frequency distribution regarding the differential direction value for the inspection region M ₂ , the magnitudes of the total sum γ, δ of the frequency for the specific range (two poles) of the differential direction value and the predetermined threshold values SL4, SL5 are compared. . Here, when both γ> SL4 δ> SL5 are satisfied, it is determined that the welding state is bad, and when neither is satisfied, it is determined that the welding state is good. Here, even if the welding state is good, specular reflection light may be generated, but the position where specular reflection light is generated is displaced and the frequency distribution of the differential direction value is also different. Can be identified. The other procedures are the same as those in the first embodiment, and the description thereof will be omitted.

(Embodiment 3) This embodiment is a combination of the methods of Embodiments 1 and 2, and has an inspection area M
_{By making the} judgment according to ₁ and the judgment according to the inspection region M ₂ , the quality judgment can be made more accurately. Here, except when both are judged to be good,
If it is judged as defective, the reliability of non-defective products is improved.

[0021]

According to the first aspect of the present invention, since the frequency distribution regarding the differential direction value is obtained for the pixels on the boundary line of the unevenness formed by laser welding and compared with the frequency distribution of the non-defective product, the light source and the image input device are combined. The positional relationship does not need to include the height information of the surface of the inspection object, and may be set so that a shadow corresponding to the unevenness of the welding location is formed. That is, there is no need for a special light source for irradiating a band-shaped parallel light beam, which is advantageous in that the structure of the light source is simplified and the arrangement of the light sources is reduced. Also, with regard to image processing, there is no need to perform complicated calculations that include information about the height of the surface of the inspection target. In addition to general processing such as differential processing and thinning processing, addition and simple comparison of size comparison Since it suffices to perform such processing, there is an effect that the processing process of image processing is simplified.

In the methods of claims 2 and 3, the quality of the judgment can be easily made by devising the method of setting the inspection area, and the frequency distribution regarding the differential direction value of the inspection object and the frequency of the non-defective product. Compare with the distribution,
Since the range of the differential direction value is limited and the frequency of the range is compared with the threshold value, it is performed by a simple process.
This has the effect of further simplifying the process of image processing for determining the quality of the welded portion.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing a procedure of a first embodiment.

FIG. 2 is an explanatory diagram showing a state of reflected light on an inspection object before welding in the example.

FIG. 3 is an explanatory diagram showing a state of reflected light on an inspection object after welding in the embodiment.

FIG. 4 is a perspective view showing a welded state of a workpiece in the example.

FIG. 5 is a diagram showing an example of setting an inspection area in the first embodiment.

FIG. 6 is a diagram illustrating a concept of raster scanning according to an embodiment.

7A and 7B show examples of frequency distributions regarding differential direction values in Example 1, where FIG. 7A shows a case where the welding state is good, and FIG. 7B shows a case where the welding state is bad.

FIG. 8 is a diagram showing an example of setting an inspection area in the second embodiment.

FIG. 9 is a block diagram of a processing circuit used in the embodiment.

10A and 10B are views showing the concept of a local parallel window in the embodiment, FIG. 10A is a position in an image, and FIG. 10B is a diagram showing a pixel configuration of the local parallel window.

FIG. 11 is a diagram showing a concept of a differential absolute value image in the example.

FIG. 12 is an explanatory diagram of a process of obtaining an edge image from a differential absolute value image in the embodiment.

FIG. 13 is a diagram showing a relationship among an original image, a differential absolute value image, a differential direction value image, and an edge image in the example.

[Explanation of symbols]

 1 Image Input Device 2 Analog-Digital Converter 3 Preprocessor 4 Original Image Memory 5 Differential Absolute Value Image Memory 6 Differential Direction Value Image Memory 7 Edge Image Memory 8 Arithmetic Processor

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI Technical display location G01N 21/88 Z 7172-2J H04N 7/18 B

Claims (3)

[Claims] 1. A pass / fail judgment of a welding state is performed by imaging an image of a spatial region including a welded portion of an inspection object welded by laser welding in which a plurality of workpieces are superposed and welded, and then image processing is performed. In the welding state determination method by the visual inspection to be performed, based on the original image which is a grayscale image input from the image input device, the differential absolute value corresponding to the change rate of the density with the adjacent pixel is set as the value of each pixel. A value image, a differential direction value image in which the differential direction value corresponding to the changing direction of the density of adjacent pixels is used as the value of each pixel, and an edge image indicating the boundary line of the density change are created, and the inspection including the welding point is made. After setting the area, the pixel on the boundary line in the edge image whose differential absolute value is larger than a predetermined threshold is found as a welding feature pixel in the inspection area, and the frequency related to the differential direction value of the welding feature pixel is obtained. A welding state determination method by visual inspection, characterized by determining the quality of the welding state by comparing the distribution with a preset frequency distribution of non-defective products. 2. The inspection region is set so as to include a welded portion and a non-welded portion on one of the workpieces, and the sum of frequencies in a predetermined range of the differential direction value for the welding feature pixel is set. The welding state determining method according to claim 1, wherein the welding state is determined to be good when the sum of the frequencies of non-defective products in the same range is larger than a preset threshold value. 3. The inspection area is set inside the welding point,
It is characterized in that the welding state is determined to be defective when the sum of the frequencies of the differential direction values in the predetermined range of the welding feature pixels is larger than a threshold value set in advance as the sum of the frequencies of non-defective products in the same range. Item 1. A method for determining a welding state by the visual inspection according to Item 1.