JPH063123A - Method and equipment for visual inspection - Google Patents

Abstract

PURPOSE:To obtain a method and equipment for visual inspection which make it possible to determine the quality of an object of inspection, accurately and in an optimum state at all times. CONSTITUTION:A plurality of light sources 11 to 17 disposed above an object 3 of inspection are used, the object of inspection is illuminated from different directions sequentially by these light sources and images thereof are picked up by image pickup cameras 21 to 25 in the sequential illuminations, whereby a plurality of image data are obtained. Meanwhile, slit light illumination is conducted by using slit light generating parts 31 and 32 and images are picked up at a prescribed angle from above, whereby optically cut image data are obtained. By using these data together, the state of the object of inspection is recognized and determined. According to this constitution, the content of determination can be substantial, performances of levels of inspection and measurement can be improved sharply and the efficiency of inspection of a soldered state can be improved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inspection method and an inspection apparatus for determining whether a mounting state of an object is good or bad based on its appearance, and particularly to a soldering portion of an electronic component after mounting on a substrate and an electronic component. The present invention relates to a visual inspection method and a visual inspection apparatus suitable for inspecting a mounting state as a target to be inspected for the quality of a mounting state based on the visual state.

[0002]

2. Description of the Related Art As a conventional technique of this type, for example, a device shown in FIG. 2 is known. And with this device,
For example, four light sources 11, 13, 15, and 17 composed of annular lamps having different diameters, and above the light sources 11 to 11 are installed above the inspection object 3 on the central axis of the light sources. Using the image pickup camera 21, the soldering portions of the inspection object 3 on the mounting board (circuit board) 2 horizontally mounted on the mechanism section 1 moving in the XY directions are sequentially moved in different directions (angles). ), The reflected light obtained from the soldering surface each time is converted into an electric signal by the imaging camera 21, and image information of a plurality of inclined surfaces of the soldering surface is obtained. , The soldering state based on each judgment level, for example, good, insufficient solder,
Judgments were made by classifying into excessive solder, no solder, and lead shift.

In addition to the above, as an apparatus of this type, for example, an apparatus described in JP-A-60-154143 can be cited.

[0004]

By the way, recently, for the purpose of downsizing of electronic equipment, high density mounting and surface mounting of substrate have been rapidly progressed, and ultra-small chips and fine pitch I have been developed.
With the mounting of C, the size of the soldering part is miniaturized and the number of soldering points is significantly increased. Therefore, the soldering quality can be detected and grasped at a high speed at high speed, and the judgment management can be performed. Becomes a very important issue.

Therefore, it is necessary to analyze the shape states of soldering and mounting with higher accuracy, accurately classify and grasp them, and make a recognition judgment, but in parallel with this, at this time, It is necessary to improve the recognition performance, that is, the defect detection rate, and to improve the false alarm rate in which a non-defective product is determined to be defective. Therefore, it is necessary to further optimize the identification and increase the speed.

However, in the above-mentioned conventional technique, it cannot be said that the recognition technique for classifying and determining the soldering state is sufficiently dealt with, and there is a problem in that the correspondence is limited. .

The present invention has been made in view of the above circumstances, and its purpose is to determine the shape of the soldered portion to be inspected, and the posture and mounting position of the electronic component in the mounted state. It is possible to clearly recognize whether or not there is, and thereby it is possible to clearly discriminate whether the inspection object is a good product or a defective product, and for the remaining part that cannot be clearly discriminated as a good product or a defective product, A visual inspection method capable of accurately and optimally determining the state of the shape of the ambiguous soldered portion and the mounting state by extracting the shape feature again in detail, and To provide the equipment.

[0008]

In order to achieve this object, the present invention uses a plurality of light sources arranged above an object to be inspected, and these light sources illuminate the object to be inspected successively from different directions. Each time, the inspection target is imaged to obtain a plurality of image data, and the size and shape of the inspection target are obtained by performing arithmetic processing on these image data for each pixel,
Furthermore, from the data representing this shape, the features of the pattern are extracted to determine whether the pattern is good or defective, and
In parallel with this, using a plurality of slit light sources arranged above the inspection object, slit light irradiation is performed on the inspection object, and a predetermined angle is set to capture an image from above to obtain light section image data. , This optical cutting image data is subjected to image processing to obtain shape line data, the position and shape of the inspection object are known from this, and a good product or a defective product is determined. The state is recognized and determined.

[0009]

Operation: By illuminating a certain portion of the object to be inspected at different angles, the reflected light peculiar to the angle is obtained. This reflected light is imaged from above and converted into an electrical signal. The image distribution data of the upward reflected light from each illumination angle obtained in this way is arithmetically processed for each pixel to obtain detailed image information. Based on this result, shape inclination angle data, shape height data, shape length data, shape area data, etc. are obtained,
The size and shape of the object to be inspected are obtained, the characteristics of the shape data pattern are extracted, and good and defective products with clear soldering portions are extracted and determined.

On the other hand, at least for the vague / indefinite determination mounted component, the slit light is radiated to the inspection object by the illumination means arranged above the inspection object to generate slit light. The reflected light is imaged from diagonally above with a certain angle as a light cutting line and converted into an electric signal. Each photographed image of the light cutting line thus obtained is subjected to image processing to obtain shape line data. Based on this result, the shape line data is compared with the reference shape line data of the inspection target object for shape comparison and classification, and good and defective products are extracted and determined. The recognition is determined.

This will be described in more detail based on an embodiment as follows. Illumination is performed by changing the angle with respect to the soldering surface that is the inspection object. Reflected light peculiar to each angle is generated from this soldering surface. The reflected light is imaged from above to obtain image information of each tilt angle of the soldering surface, and then the detailed image information of the soldering state is converted into data for each pixel. These data are divided into shape inclination angle data of the row direction component of the soldered portion, shape height data, shape area data, shape inclination angle data of the column direction component parallel to the lead tip, and further shape height data. Has been.

Next, with respect to the lead shape and the component mounting shape of the soldered portion, the slit light illuminating means arranged above the inspection object irradiates the inspection object with light and reflects it. The light is cut at a certain angle as a light cutting line, and is imaged from diagonally above, and is converted into an electric signal. Each captured image of the reflected light obtained in this way is subjected to image processing to obtain convex or concave shape line data for each inspection object. Here, especially above the inspection object,
For example, the image pickup means such as a video camera mainly functions to confirm the position of the component, while the image pickup means diagonally above is mainly used to determine the mirror-like soldering state of the soldered portion of the component. To do.

As the identification means, first, as a size comparison, a comparison is made by calculating shape values such as a solder height, a solder length, a solder area, and a solder inclination angle. The shape data is used as the feature extraction of the data pattern.
The sharpness of the array shape inclination angle and height data, the presence or absence of a relatively flat part at the center of the soldering and its size, the numerical symmetry comparison of the entire array, And the comparison of the amount of symmetry deviation, the comparison of the patterns of equiangular lines, and the extraction and creation of the features of the patterns of other shape data. This method / means has a feature that it is easy to grasp the state and shape pattern of the mirror surface of the solder surface for every pixel.

Thereby, for example, as typical judgment results showing the soldering state, (1) good, (2) wetting failure,
(3) Insufficiency, (4) No solder, (5) Excessive solder, etc., and the indirect component mounting state determination results are (6) Lead deviation, (7) Lead floating, (8) Extraction judgment results such as lead not attached, (9) no component, (10) component misalignment, (11) component floating / part standing, etc. are obtained.

Next, the height of the convex or concave shape line and the height of the convex or concave shape line compared with the reference shape line data of the inspection object are mainly based on the mounting state of the parts. The amount of misalignment between left and right and up and down, and the size of uplift can be grasped. This way
The means has a feature that it is easy to grasp the shape in the vertical direction.

Therefore, by grasping these information, it is possible to cope with subtle changes in soldering conditions and variations, and in particular, to make a determination of the joint state and soldering state of the soldered portion and the lead state and the component mounting state. . From this, for example, as the judgment result of the soldering state and the mounting shape state of the component,
(6) Lead misalignment, (7) Lead floating, (8) Lead not attached,
The extraction judgments such as (9) no component, (10) component shift, and (11) component floating / part standing are obtained.

In the inspection, it is absolutely necessary to avoid overlooking defects, and it is necessary to minimize misperceptions of non-defective products. Therefore, in the present invention, these both means are used in combination,
The shape recognition judgment of the right material and the right place is performed. That is, after performing the typical extraction judgment by the previous means,
As for the method of determining the latter means, at least only the ambiguous / indeterminate non-defective product is finally determined to confirm the mounting state of the non-defective product.

According to this method, the contents of judgment can be enhanced, the performance at the inspection and measurement levels can be greatly improved, and the inspection efficiency of the soldering state can be improved.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The appearance inspection method and the appearance inspection apparatus according to the present invention will be described in detail below with reference to the illustrated embodiments.
1 and 3 show an embodiment of the present invention. In these figures, 4 is a lighting sequential switching unit, 10 and 10 'are imaging camera switching units, 8 is a recognition determination unit (CPU), and 31 and 32 are Slit light generation unit, 33 slit light generation switching unit, 60 bus line, 61, 61 ', 61 "I / O, 62 operation unit,
63 is an image interface, 64 is an image monitor, and 65
Is an external memory interface, 66 is an FDD (floppy disk device), 67 is an HDD (hard disk device), 68 is a data display device, 69 is a mechanism control unit, and other symbols will be sequentially described below.

Above the inspection object 3 of the mounting board 2 mounted on the mechanism section 1 which moves in the XY directions, the light source 11, the light source 13, the light source 15, and the light source 1 are sequentially arranged from above the mounting board 2.
7 are arranged in four stages to form a step illumination unit. Each of the light sources 11, 13, 15, and 17 is formed of a ring arranged concentrically, and has a diameter that gradually increases toward the mounting substrate 2 and is sequentially switched by the illumination sequential switching unit 4. The mounting board 2 is illuminated.

Further, from above the light source 11, there is one image pickup camera 21 of the image pickup section on the central axis of the step illumination section and directly above the inspection object 3, and diagonally upward from both sides in the X axis direction from the central axis. 2 image pickup cameras 22 and 23, and 2 image pickup cameras 24 and 25 in the Y-axis direction (24 and 25 are not shown in the figure), a total of 4 image pickup cameras are symmetrical in 4 directions. It is arranged in a way. Then, the image pickup camera 21 and any one of the image pickup cameras 22, 23, 24, and 25 are connected to the light source 1.
Images of reflected light of the inspection target object 3 illuminated by sequentially switching 1, 13, 15, and 17 can be sequentially captured.

These image pickup cameras 21, 22, 23,
The output signal of either 24 or 25 is output from the imaging camera switching device 1
Image memory unit A5, B5 'of RAM through 0, 10'
Therefore, the image memory units A5 and B5 'are connected to the light sources 11, 13, and 1, respectively.
Images are stored in correspondence with 5 and 17. At this time, the light source 1
The images of the inspection object 3 illuminated by the sequential switching of 1, 13, 15, and 17, the image signals of the imaging camera 21 are stored in the image memory A5, and the imaging camera 22,
The image signal of any one of 23, 24 and 25 is stored in the image memory B.
5'is input sequentially.

Next, the outputs from the image memories A5 and B5 'are input to the image calculation processing section 6 for each pixel. In the image calculation processing unit 6, the image pickup data corresponding to the illumination from each of the light sources 11, 13, 15, and 17 are compared with each other for each image pickup pixel, and the magnitude of the illuminance is relatively compared. From the data, the captured image with the highest image illuminance level is classified and selected. The selected data is input to the shape code data pattern unit 7, and the image number code indicating the angle of the reflection surface of the picked-up image selected here is organized as shape inclination angle data.

The recognition determining unit 8 is composed of a CPU and has a shape
Based on the coded data from the data pattern section 7, the soldering state is divided into its shape, height, length, area, etc., and the respective sizes are calculated and calculated. Furthermore, the recognition determining section 8 can accurately determine whether or not the soldering section formed on the surface of the inspection object 3 of the mounting board 2 is formed in a desired state. There is.

Next, the operation of this embodiment will be described. FIG. 5 is a cross-sectional view showing the details of the soldering portion 51 of the mounting board 2, and the diagonally upper imaging cameras 22, 23, 24, and
Alternatively, the reflected light incident on any one of 25 is shown, but at this time, the inspection object 3 is also an object of shape state determination. In FIG. 5, a copper foil pattern 52 is formed on the main surface of the mounting substrate 2, and a lead portion 53 (lead portion of an electronic component) is connected to the copper foil pattern 52 via a soldering portion 51 so as to be connected to the copper foil pattern 52. It is fixed. Further, in the figure ,, is the data number of the code of the shape inclination angle, and the arrow indicates the direction of the reflected light.

In the mounting substrate 2 having such a structure, after the light from the light source 11 is reflected on the surface of the soldering portion 51, the light from the light source 13 is reflected on the surface of the soldering portion 51 in the direction of the data number. Then, after the light from the light source 15 is reflected on the surface of the soldering portion 51 in the direction of the data number, in the direction of the data number and after the light from the light source 17 is reflected on the surface of the soldering portion 51, the data number of Each of them advances in the direction, and is incident on any of the imaging cameras 22, 23, 24, or 25.

As described above, in each image pickup data corresponding to each illumination, the image calculation processing unit 6 of the image signal classifies the image pickup image having the highest image illuminance level, and the code of the shape inclination angle of the selection target image is obtained. Set relatively for each pixel,
In the shape code data pattern section 7, this set data is coded and organized. Therefore, in the shape code data pattern unit 7, as shown in the plan view of FIG. 6, the data number having the maximum illuminance in each pixel is written in the corresponding pixel in the image. The data will be organized.

Therefore, this data is a detailed data state diagram showing from the data number of the code of the shape inclination angle from the upper surface of the soldering. Here, they are arranged as data of the shape inclination angle code, and are stored, recorded, displayed, and output.

Next, the data number of each pixel,
A method for determining ,,,,, will be described below. Note that, as shown in FIG. 5, only the data numbers, ... Are directly obtained in this embodiment, and the rest are obtained by interpolation. First,
In order to obtain an accurate angle, if data is obtained using a spherical body having a mirror-like surface similar to the reflectance of the soldered portion 21, an analog data distribution state of illuminance can be obtained. At this time, for example, the illuminance is taken with 256 gradations,
The lights are spaced at approximately equal angles. This is used as the data of the reflection code on the surface of the soldering portion 51 for illuminating each stage, and corresponds to the data number by the light source 11, the data number by the light source 13, the data number by the light source 15, and the data number by the light source 17. Number them.

Next, since the curvature of the surface of the soldering portion 51 is continuous, the data number between the light source 11 and the light source 13 at each stage is calculated by the interpolation method of the intermediate analog value.
The data number can be set between the light source 13 and the light source 15, the data number can be set between the light source 15 and the light source 17, the data numbers other than the light source 17 can be set, the data numbers other than the light source 11 can be set, and the like.

As a concrete method of such interpolation, for example, when the level is 100 gradations or less,
With this, the part where the illuminance of the data number and the illuminance area of the data number overlap is taken as the data number, and similarly,
The data number, the data number, and the direction of the data number may be set. Therefore, in this case, the solder surface observable range can be classified into 9 levels of angles depending on the board component of the solder surface.

FIG. 6 shows the obliquely upper imaging cameras 22, 23,
In the figure showing the state in which the data number for each pixel of the mounting substrate 2 on which the soldering portion 51 is formed is written by 24 and 25, the light source 11 of the step illumination portion having four stages,
The imaging cameras 22, 23, 2 that capture the reflected light from the soldering portion 51 when 13, 15, 17 are sequentially switched.
The image calculation processing unit 6 takes a photographed image from any one of Nos. 4 and 25, and determines the region of the portion having a relatively high reflectance from the surface of the soldering unit 51 as each of the light sources 11, 13, 15, and 1.
7 shows the extracted state corresponding to 7.

In FIG. 6, the outer frame of the rectangle is photographed by one of the image pickup cameras 22, 23, 24 and 25, and
The area to be inspected in the relevant portion is shown. A region with a data number in this range is a portion of the reflected light from the light source 11 having a relatively high illuminance. A region with a data number is a light source. A portion of the reflected light from 13 having a relatively high illuminance, a region with a data number is a portion of the reflected light from the light source 15 with a relatively high illuminance, and a region with a data number is from the light source 17. The relatively high illuminance of the reflected light is shown. Also, the data number ...
The areas marked with are calculated and set by the above-described interpolation method.

The position of the end portion of the lead portion 53 shown in FIG. 6 is set in advance by the image pickup of the upper image pickup camera 21. The result shown in FIG. 6 shows that the pixel data of the detailed data state diagram of the shape information (after thinning as needed when there is a lot of redundancy) is row data n × column data m = nm (vertical X horizontal), knitted in the shape code data pattern portion 7 and stored.

With such a distribution (of the reflected light having a relatively high illuminance), a change in the inclination angle of the surface of the soldering portion 51 can be recognized according to the distribution state. That is, as shown in FIG. 7, when the angle of the light incident on the surface of the soldering portion 51 from the light source with respect to the vertical line is α and the angle with respect to the vertical line of the camera optical axis on the solder surface is β, the reflected light This is because there is a relation that θ = α + (β−α) / 2 with respect to the inclination angle θ of the portion (region) having high strength.

Therefore, by this, the light source 1 of the step illumination unit is
It is possible to obtain the tilt angle of a portion in a region where the reflected light from any one of the illumination units is relatively strong among 1, 13, 15, and 17. In this embodiment, the above-mentioned code is used for data conversion, and the height of each portion on the soldering surface is calculated.

FIG. 7 shows the image pickup cameras 22 and 2 obliquely above.
FIG. 3 is a diagram showing a side surface of a soldering portion showing reflected light incident on any one of 3, 24, and 25, and a method for obtaining the height of each portion of the soldering surface will be described below based on these drawings. .

In the shape code data pattern portion 7, as shown in FIG. 7, the inclination angle of the solder surface shape is θ and the vertical axis of the light incident from the step illuminating portion corresponds to each pixel. Let α be the angle, β be the angle from the vertical axis of the imaging optical axis of the obliquely upper imaging camera 22 to, and let a be the length of a unit pixel, then the following equation holds: hn≈ {a / cos (β −θ)} × sin θ = [a / cos {(β−α) / 2}] × sin {α + (β−α) / 2} The solder shape height hn for each unit pixel length a
(n = 1, 2, ...) And HN = HM + hn is obtained as cumulative shape height data.

Therefore, at this time, the obtained numerical value is divided into relative numerical values as a parameter for feature extraction and is obtained as data, so that the array of data for each pixel can be utilized. In addition, since the height is obtained corresponding to each pixel in this way, each pixel length component [{a /
The cross-sectional area of the solder on a certain side surface can be calculated by the product of cos (β−θ)} × cos θ]. In addition, the overall soldering area, height above a certain level,
Alternatively, as the area occupied by the solder shape at an angle, it is possible to obtain the number of pixels × the vertical / horizontal length (a / b) of the pixels.

Next, a shape data representing the shape of soldering.
Determine the pattern. One of the properties representing the shape of soldering is steepness, which is shown in FIG.
In the explanatory view showing the side surface of the soldering part, the shape of the side surface of the Nth row is determined by whether it is a straight line, a convex line, or a concave line. When the difference for each pixel is A = (N)-(M), most of the shape inclination angle data have a linear change when A = 0, and a concave line change when A> 0. When the sex is A <0, it can be judged that the convex line change is slow. Note that the rows and columns here mean the rows and columns of pixels, as shown in FIG.

The recognition determination unit 8 is configured to compare each identification item obtained by each of the above-described calculations with a reference value for determination.

Next, the classification items necessary for the discrimination judgment are set. Here, as the classification items, for example, (1) good, (2) poor wetting, (3) so that these items can sufficiently cope with subtle changes in soldering conditions and variations.
Insufficient, (4) No solder, (5) Over solder, (6) Lead misalignment, (7) Lead floating, (8) Lead not attached, (9) No component,
(10) Parts misalignment, (11) Parts floating / part standing, etc. are set.

In the case of automatically classifying each of the classification items according to the present invention, as an example of each shape of the soldering state, in particular, in this embodiment, the soldering state of the soldered portion and the lead are used. It is adapted to judge the joining state of the state.

The pattern of the shape code information data in the soldering state in this embodiment includes the height of the solder,
It includes shape data such as cross-sectional area, length, real area, steepness, symmetry, and central flatness (discontinuity) of special shape.
It should be noted that these are stacked from an array of data on the shape of the solder surface.

In this embodiment, by combining the above-mentioned identification items and extracting features, the inspection object is
Which of these classification items is applicable is determined. Explaining in detail, as a concrete example of the feature code extraction by shape code information data of the soldering state and the pattern comparison, first, the contents to be directly judged what the soldering state is as follows. Become.

(A) Features of solder height, length, area, and inclination angle as standard values, large steepness (concave line change), no central flat area, and lead end detection When is extracted. In this case, the classification items → (1) Good solder.

(B) The solder height is very low, the length is very short, the area is very small, and the inclination angle is a little small value.
When each of the features of large steepness (change in concave line), no central flat area, and lead end detection is extracted. In this case, classification items → (2) Solder shortage.

(C) Solder height is very low, length and area are normal, inclination angle is very small, slow (convex line change), central flat area is rather large, and lead edge is With detection,
When each feature of is extracted. In this case, classification item → (3) Insufficient solder wetting.

(D) Solder height, length, area, inclination angle,
When each feature of steepness, all values are null, and lead end is detected is extracted. In this case, classification items → (4) No solder.

(E) Solder height is very high, length is very long, area is very large, slow (convex line change), central flat area is large, and lead end is not detected. When each feature is extracted. In this case, classification item → (5) Excessive solder.

Next, in addition to the above, the classification items to be judged indirectly depending on the soldering state are as follows in indirectly judging the mounting state of the parts.

(F) When each feature of the shape of the solder is asymmetric and non-uniform in the tip shape, and there is a deviation in the detection of the lead end. In this case, classification item → (6) Lead position is divided.

(G) Solder height, length, area, and inclination angle are larger than the standard, slow (convex line change), no central flat area, and solder enters from the lead end. When the feature is extracted. In this case, classification item → (7) Lead floating.

(H) The height of the solder is usually very long, the area is very large, the matrix ratio is very large, the inclination angle is very small, and there is slowness (convex line change). , The area of the central flat part is a little larger, and the solder gets in from the lead end. In this case, category item → (8) Lead has not arrived.

(I) The height of the solder is usually very long, the area is very large, the amount is very large, the matrix ratio is very large, the inclination angle is very small, and the Convex line change), the central flat area is a little larger, and the lead end is not detected. In this case, classification items → (9) No parts.

(J) When the characteristics of the solder are extracted such that the tip shape is asymmetrical and non-uniform, and the detection of the component electrode end is deviated. In this case, the classification item → (10) Parts position is divided.

(K) Solder height is usually long, length is very large, area is very large, matrix ratio is very large, inclination angle is very small, and there is slowness (convex line change). , The area of the central flat part is slightly large, and the solder penetrates into the parts from the ends of the terminals. In this case, classification items → (11) Standing parts.

Therefore, in order to accurately obtain these classifications,
In this embodiment, the structure described below is provided. Returning to FIG. 1 and FIG. 3, in these drawings, above the inspection object 3 on the mounting substrate 2, there are a slit light generating section 31 in the X direction, a slit light generating section 32 in the Y direction, and a reflection mirror in the X direction. -Angular variable mechanism 34 and Y-direction reflection mirror-
An angle changing mechanism 35 is arranged. 36 is a reflection mirror angle control section.

Here, the slit light generator 31 in the X direction and the slit light generator 32 in the Y direction are actually arranged at right angles to each other in the X axis direction and the Y axis direction (in FIG. 1, it is easy to understand). Therefore, the slit light generator 32 in the X direction is
Direction on the opposite side of the slit light generator 31).
Then, switching between the slit light generation units 31 and 32 is performed by the slit light generation switching unit 33.

Next, the operation of this embodiment will be described. First, the slit light generator 31 in the X direction, which is arranged in the X-axis direction above the inspection object 3, generates vertical slit light, The reflection mirror angle control unit 36 controls the X-direction reflection mirror angle changing mechanism 34 to sequentially irradiate the slit light in the X-axis direction of the inspection object 3 while changing the reflection mirror angle. This irradiation light becomes a light cutting line with respect to the inspection object 3, and the images thereby are sequentially picked up by the image pickup camera 22 or 23 in the X-axis direction obliquely above at a certain angle. There is.

Here, the image pickup camera 22 or 23 is selected such that a part of the inspection object 3 is in the light ray shadow of the inspection object 3 itself or another adjacent inspection object 3'when the imaging direction by these is selected. It is done so that there is no direction.

Next, a lateral slit light is generated by the slit light generator 32 in the Y direction arranged in the Y-axis direction above the inspection object 3, and a reflective mirror angle controller 36 causes a reflective mirror angle. While controlling the variable mechanism Y35 and changing the reflection mirror angle thereof, horizontal slit light is sequentially emitted in the Y-axis direction of the inspection object 3. This irradiation light becomes a light cutting line with respect to the inspection object 3, and an image by this is taken at an obliquely upper angle in the Y-axis direction by the image pickup camera 24 (disposed in front of the front in the figure) or imaged. Images are sequentially captured by the camera 24 (arranged on the front rear side).

Here again, the selection of the imaging cameras 24 and 25 is such that the imaging directions by these do not cause a part of the inspection object 3 to fall within the ray shadow of the inspection object 3 itself or another adjacent inspection object 3 '. It is performed in the direction.

FIG. 8 shows the image pickup operation by the light section line at this time, and FIG. 8A shows the vertical light section line 1-1 'by the slit light generator 31 in the X direction. When the reflected light is imaged by the imaging camera 22, (b) irradiates the inspection object 3 with a vertical light cutting line 2-2 ′ by the slit light generation unit 31 in the X direction. When the reflected light is imaged by the imaging camera 23, (c)
Is a case where the horizontal light cutting line 3-3 ′ is applied to the inspection object 3 by the Y-direction slit light generator 32 and the reflected light is imaged by the imaging camera 24, and (d) is the light It shows a case where the vertical light cutting line 4-4 ′ is irradiated to the inspection object 3 by the cutting line generating unit Y32 and the reflected light is imaged by the imaging camera 25.

As a result, the light cutting line is formed into a convex shape or a concave shape depending on the image pickup direction, depending on the shape of the image pickup target portion. In this way, the image by the optical cutting line imaged by the imaging cameras 22 to 25 is processed in the image processing unit 71, and compared with the image without the optical cutting line in advance here, the background is canceled from these two types of images, and Only the image data by the cutting line is taken out, and the result is the shape line data section 72, which is the shape of the convex or concave optical cutting line obtained corresponding to the inspection object 3. It is formed into line data. Further, the shape line data section 72 has a standard serving as a standard for component mounting in accordance with a preset classification such as height, length, width, etc. for each position, type and shape of each inspection object 3. Shape line data is set and registered.

Therefore, the recognition determining unit 8 stereoscopically recognizes the position and shape of the component based on the shape line data and the reference shape line data, and recognizes the shape line data and the reference shape line data. As a result of comparing the above, it is determined as follows.

When the amount of deviation from the reference line to the left and right exceeds a specified value Classification item → (6) Lead deviation When the amount of deviation from the reference line above and below exceeds the specified value Classification item → (7) Read Floating, or (8) No lead is attached.If a convex or concave optical cutting line cannot be obtained with respect to the reference line.Category item → (9) No parts. Case Classification item → (10) Component deviation When deviation amount above and below the reference line exceeds a predetermined value Classification item → (11) Component standing An example of the above determination process will be described with a specific example.

First, in the case of non-defective products with leads shown in FIG. 9A and in the non-defective products with parts shown in FIG. 9C, shape pattern data: shape of solder center → reference state shape pattern Data: Shapes on both sides of solder → reference state Shape line data: Height of lead or component → reference value. Therefore, the code distribution at this time is, for example, as shown in FIG.
As shown in 0.

On the other hand, in the case of the defective product with the lead shown in FIG. 9B and the defective product with the component shown in FIG. 9D, the shape pattern data: the shape of the center of the solder → the reference state Shape pattern data: Shapes on both sides of solder → steep state Shape line data: Height of lead or component → Higher than standard value. Therefore, the code distribution at this time is, for example, as shown in FIG.
Therefore, the quality can be determined from these code distributions.

Therefore, the identification determination operation according to the above embodiment is shown in a flow chart of FIG. First, an image is captured by the image pickup camera 21 located directly above, thereby confirming the component position of the inspection object 3 and the entire position of the lead in the image capturing range. An image is picked up by any one of 25 and the entire lead position of the image capturing range is confirmed (S1 to S3,
(S stands for step).

Next, based on these position confirmation results, the inspection regions necessary for checking the soldering state for each lead are set as windows, and the image in the window is already described for each window. By the above method, a shape code is formed for each pixel, and a shape code data pattern is created for each window (S4 to S6). At this time, in S6, in order to characterize and determine the soldering state, each identification item of height, length, area, inclination angle, steepness, and central flat area is determined from the shape information data of the soldering state. The shape absolute value or relative value, and the shape pattern etc. are created and classified.

From this, as typical classification states showing the soldering state, (1) good, (2) poor wetting, (3) insufficient,
(4) No solder, (5) Excess, (6) Lead deviation, (7) Lead floating, (8) Lead not attached, (9) No parts, (10) Parts deviation, (11) Parts standing, etc. The extraction determination is performed, and the non-defective product is extracted in S7 and the defective product is extracted in S8.

However, in reality, there are inspection objects that cannot be clearly determined in either S7 or S8. Therefore, with respect to the inspection object that cannot be definitely determined as a defective product, but on the other hand, even if it is a non-defective product, only an ambiguous / indeterminate determination can be made, the process proceeds to S9 and subsequent steps. It classifies which item such ambiguous / indeterminate judgment has appeared (S9 to S).
14).

Next, an image is picked up by the image pickup camera 21 directly above and the position of the component of the inspection object is confirmed (S
15), while confirming the entire position of the lead within the image capturing range, further confirm the entire position of the lead within the image capturing range again by the image data obtained by any one of the obliquely upper and upper imaging cameras 22 to 25. (S16) Then, based on these results, the window for defining the inspection area necessary for checking the lead and soldering state is reset (S16).
17).

Next, slit light is generated by either of the slit light generators 31 and 32, and the window of the set inspection region is irradiated with a light cutting line by the reflection mirror angle control (S18). Is imaged by any of the obliquely upward and upward imaging cameras 22 to 25 (S19),
From this imaged data, the shape line data of the convex or concave light cutting line is obtained (S20).

In this way, if the shape line data of the convex or concave light cutting line is obtained, the position and shape of the component can be recognized three-dimensionally from this.
Therefore, as a result of comparing the shape line data with the reference shape line data, the following determination processes are sequentially performed in S21 to S25.

S21 When the amount of left / right deviation from the reference line exceeds a predetermined value: Classification item → (6) Lead deviation S22: When the amount of vertical deviation from the reference line exceeds a predetermined value: Classification item → (7) Re -Floating, or (8) Lead not attached S23 When a convex or concave optical cutting line is not obtained with respect to the reference line Classification item → (9) No parts S24 A certain amount of deviation from the reference line is specified. If it exceeds the value Classification item → (10) Parts deviation S25 If the amount of deviation above or below the reference line exceeds the specified value Classification item → (11) Parts standing In this case, of course, the shape line of the optical cutting line If there is an abnormality in the data, it is defective, and if there is no abnormality, it is a non-defective product, and this is judged in S20 to finally determine whether it is a non-defective product.

Therefore, according to this embodiment, when only an ambiguous or uncertain decision is made as to whether it is a non-defective product, a final decision as to whether it is a non-defective product or a defective product is made depending on the mounting state of the components. Therefore, it is possible to always reliably and accurately obtain an accurate quality judgment.

By the way, as described above, in such a case, it is absolutely necessary to avoid overlooking defective products. Therefore, increasing the defect detection rate is a major proposition, and similarly, misidentification of non-defective products is minimized. Therefore, lowering the false product false positive rate is also a major proposition.

Therefore, in this embodiment, as shown in the flow chart of FIG. 4, two types of judging means are used in combination to perform shape recognition judgment of the right material in the right place. For ambiguous boundary values or unclear shapes, the first means for determining the shape data and pattern by the image pickup means consisting of the step illumination part and the diagonally upper image pickup camera 21, the slit light and the diagonally upper image pickup. A second technique based on the determination of the optical cutting line by the image pickup device including the cameras 22 to 25, and a determination technique based on an AND condition that finally determines a non-defective product only when it is determined to be a non-defective product by either device;
If any of the results of the determination by the first means and the determination by the second means do not match, it is not determined as a non-defective product, and a determination technique by an OR condition that gives priority to the case where either one of the determinations is not a good product Therefore, a reliable judgment result can always be obtained.

Therefore, according to this embodiment, after the typical determination of the inspection object is first obtained, the ambiguous / uncertain boundary value or pattern of the remaining inspection object can also be accurately determined. Therefore, the determination content can be enhanced, the efficiency of inspection and measurement can be significantly improved, and the soldering state inspection and measurement level can be made accurate.

At this time, depending on the identification item, depending on the identification item, the classification according to the judgment classification may not be clearly made. Therefore, an item that can sufficiently obtain the effect of the identification must be selected. Needless to say, it is also necessary to prioritize these identification items according to the clarity and importance of the identification so that feature extraction in the determination process can be performed reliably. Needless to say.

On the contrary, in each of the two judging means of the present invention, all the inspection objects are inspected by means having a high judgment priority, and in the next judging means, at least the former of the inspection objects is non-defective. Inspect only parts. With this method, the soldering state inspection and measurement level can be made accurate,
It is possible to greatly improve the efficiency of inspection and measurement and speed up the system.

That is, first, in this case, the total number of inspection objects is inspected by the shape line determination unit (that is, the second means) which is a means having a high determination priority. By the shape line determination unit, shape line data is obtained, the position and shape of the inspection object are obtained, and clear non-defective products and defects are extracted and determined, and then a part of the inspection target and a shape pattern which is another determination means. The inspection is performed by the determination unit (that is, the first means). At least, non-defective product, for the extraction determination, the shape pattern determination unit, to determine the size and shape of the inspection object, further extract the features of the shape data pattern, in particular, good soldering,
The visual inspection apparatus for soldering and mounting is characterized by extracting and determining defects and recognizing and determining the state of an inspection target together with these.

According to the above-mentioned embodiments, particularly, depending on the shape value and the shape data pattern, mainly the good solder, the insufficient solder, the poor solder wetting, the lack of solder, the excessive solder, etc. are delicate soldering conditions. Identification and judgment are performed according to changes, and lead line shapes and parts of subtle soldering parts such as lead misalignment, lead floating, lead non-attachment, no component, component misalignment, component standing, etc. are determined by shape line data. Identification and judgment can be performed according to changes in mounting shape and other conditions, and optimal and accurate judgments can be made.The judgment contents can be easily enriched, soldering state inspection, measurement level can be made accurate, and inspection, The efficiency of measurement can be significantly improved.

Further, according to the above-mentioned embodiments, the solder shape is mainly used for the shape data (height, length,
(Shape value such as area) and shape data pattern are used to identify and determine, and mainly lead and component mounting shapes are identified and determined based on the optical cutting line. Since this is done, more detailed and more accurate determination is possible.

That is, according to this embodiment, the object to be inspected is irradiated mainly with the soldering state, and the illumination means arranged above the object to irradiate the object to be inspected with different angles. The reflected light from the surface of the inspection target due to this light irradiation is imaged from above, the size and shape of the inspection target are obtained, and the characteristics of the shape data pattern are extracted to determine whether the product is a good product or a defective product. At the same time, mainly for the mounting state of the shape of the component, the slit light illuminating means arranged above the inspection object irradiates the inspection object with light, and the reflected light is used as a light cutting line to form a certain angle. Shape line data is obtained by taking an image from above and performing image processing on the captured image of the light cutting line, and the position and shape of the inspection object are obtained, and non-defective and defective products are extracted and determined In this way, the state of the inspection object is recognized and determined in combination with these, so that the inspection of the soldering state and the setting of the measurement level can be made accurately, and the efficiency of inspection and measurement can be greatly improved. .

Further, according to the above-mentioned embodiment, the illumination means arranged above the inspection object irradiates the inspection object with light at different angles, and the inspection object is irradiated with the light. The reflected light from the surface is captured from directly above and diagonally above and converted into an electric signal, and the image distribution data of the reflected light is arithmetically processed for each pixel from this electric signal to obtain the size and shape of the inspection object. In addition, the characteristics of the shape data pattern are extracted to determine whether the soldered part is a good product or a defective product, and on the other hand, for the inspection object for which only ambiguous / unclear extraction judgment is obtained, The slit light generating means arranged above irradiates the inspection object with light, picks up the reflected light as a light cutting line at a certain angle, and picks up an image from above. Image processing is performed to obtain shape line data, the position and shape of the inspection object are judged, and non-defective products and defective products are extracted and judged, and together with these, the state of the inspection object is recognized and judged. Therefore, the inspection of the soldering state and the setting of the measurement level can be accurately obtained, the efficiency of the inspection and measurement can be significantly improved, and the processing can be speeded up as a system.

Further, in the above-mentioned embodiment, the identification judgment about the soldering state of the general flat package type electronic component has been explained, but the soldering state of the J-type electronic component is also oblique. Upper imaging camera 22-
Similar to the above-described embodiment, it is easy to perform the visual inspection by the discrimination determination of the soldering state from the data imaged by any one of 25.

Further, it goes without saying that the present invention is not limited to the above-mentioned embodiment and can be applied to any object.

Further, according to the present invention, instead of the shape information data and the shape data pattern in the above-mentioned embodiment, for example, the shape inclination angle code data and the shape inclination angle code data distribution are used. It is also possible to implement a method in which features are extracted by a pattern, and thereby a clear determination of good products and defective products is made.

Further, in the above embodiment, the image pickup by the step illumination and the image pickup camera used for the image pickup by the slit light are common, but they may be implemented as separate image pickup cameras.

In the above embodiment, four light sources are used and the illumination angle is changed in four steps.
The present invention is not limited to this, and, for example, the inclination angle direction can be identified by front and back soldering illumination by halving in the angular direction of the solder surrounding annular illumination, or the solder left and right illumination can be used by quadrupling. By doing so, it is possible to improve the identification accuracy in the inclination angle direction and also enable the identification of the cut-out, excluding the influence on the adjacent parts.

It is needless to say that the present invention is not limited to four-stage illumination, and the same effect can be obtained with three stages or four or more stages.

Further, in the above-mentioned embodiment, the light emitted from the irradiation device has a single wavelength. However, it is also possible to change the color for each stage and irradiate.

That is, according to the present invention, a plurality of light sources having different hues are used, images of the soldering surface to be inspected illuminated by these are picked up from a plurality of angles, and image information of each angle of the soldering surface is obtained. To get.

As a concrete example of the present invention as described above, in the case of multi-stage illumination, in the respective stages, and in the case of multi-direction illumination of the respective directions, the respective angles, that is, the respective light irradiation angles are set. By correspondingly emitting light of a unique color,
The image distribution data can be extracted at higher speed.
For example, when the illumination means has a plurality of hues such as red, green, yellow, and blue, it is not necessary to switch or move the light source over time because each irradiation angle can be specified by the difference or change in hue of reflected light. The processing speed can be increased.

[0098]

According to the present invention, the judgment of the object to be inspected, for example, the soldered portion, can be made for each position, and the shape of the soldered state can be determined depending on the subtle changes in the soldering conditions. First of all, by using the shape data and the pattern, first of all, a distinctly good product and a defective product are discriminated and classified, and then at least each ambiguous / unclear soldering part is classified. On the other hand, since the shape line data is used to perform detailed feature extraction again on the shape of the soldered state, it is possible to easily enhance the determination content. Inspection of soldering condition. The measurement level can be made precise and the efficiency of inspection and measurement can be greatly improved.

On the contrary, in each judging means, first, the second judging means inspects all of the inspection objects, and then the first judging means judges at least the non-defective part in the former judgment. By the method of inspecting only
Inspection of soldering state, precise measurement level, inspection,
It is possible to greatly improve the measurement efficiency and speed up the system.

As a result, according to the present invention, it is possible to detect and grasp the quality of the shape state of soldering at high speed and discriminate and manage. That is, the shape state of soldering is analyzed with higher accuracy, the classification and grasping are performed accurately, the recognition determination is performed, the recognition rate, that is, the defect detection rate is improved, and a false report that is a false report that determines a good product to be defective. The rate can be improved.

[Brief description of drawings]

FIG. 1 is a simplified configuration diagram of main parts showing an embodiment of an appearance inspection apparatus according to the present invention.

FIG. 2 is a simplified configuration diagram of a main part showing a conventional example of an appearance inspection device.

FIG. 3 is a functional block diagram showing an embodiment of an appearance inspection device according to the present invention.

FIG. 4 is a flowchart for explaining the operation of the embodiment of the appearance inspection apparatus according to the present invention.

FIG. 5 is an explanatory diagram showing a method of capturing data from an inspection target object by step illumination in an embodiment of the present invention.

FIG. 6 is an explanatory diagram showing a distribution state of code data according to an embodiment of the present invention.

FIG. 7 is an explanatory side view showing a method of processing shape data according to an embodiment of the present invention.

FIG. 8 is a plan view illustrating an optical cutting line of an inspection target according to an embodiment of the present invention.

FIG. 9 is an explanatory diagram showing an example of a non-defective product and a defective product targeted by an embodiment of the present invention.

FIG. 10 is an explanatory diagram showing an example of a distribution state of code data when an inspection target is a non-defective product in one embodiment of the present invention.

FIG. 11 is an explanatory diagram showing an example of a distribution state of code data when an inspection target is a defective product in one embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Mechanism part 2 Mounting board 3 Inspection object 11, 13, 15, 17 Light source 21, 22, 23, 24, 25 Imaging camera 4 Illumination switching part 5, 5'Image memory part A, B 6 Image calculation processing part 7 Shape Code data pattern part 8 Recognition determination part 31, 32 Slit light generation part 33 Slit light generation switching part 34, 35 Reflective mirror angle variable mechanism part 36 Reflective mirror angle control part 51 Soldering part 53 reel Image part 71 Image processing part 72 Shape line data part

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] February 18, 1993

[Procedure Amendment 1]

[Document name to be corrected] Drawing

[Correction target item name] All drawings

[Correction method] Change

[Correction content]

[Figure 1]

[Fig. 2]

[Figure 6]

[Figure 3]

[Figure 4]

[Figure 5]

[Figure 7]

[Figure 8]

[Figure 9]

[Figure 10]

FIG. 11

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yasuo Nakagawa 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Inside the Hitachi, Ltd. Institute of Industrial Science (72) Inventor Yuji Takagi 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Incorporated company Hitachi, Ltd.

Claims (8)

[Claims] 1. An illumination means arranged above an inspection target object irradiates light to the inspection target object at different angles, and the reflected light from the surface of the inspection target object due to this light irradiation is applied from above. The first image is picked up, the image data of the reflected light is arithmetically processed for each pixel, the size and the shape of the inspection object are obtained, and the characteristic of the shape data pattern is extracted from this to determine the good product and the defective product. Slit light irradiation is performed on the inspection object by the determination means and the slit light generation means arranged above the inspection object, the reflected light is imaged from above as a light cutting line, and image data by the light cutting line is obtained. To obtain shape line data,
A second determination means for extracting and determining a non-defective product and a defective product by providing a position and a shape of the inspection target object is provided, and the determination results of the first and second determination means are used together to test the inspection target object. An appearance inspection method characterized by being configured to recognize and determine a state. 2. The invention according to claim 1, wherein the second determination means is applied to an inspection target object whose determination result by the first determination means is an ambiguous / unclear determination result. A visual inspection method characterized in that 3. The invention according to claim 1, wherein the first determination means is configured such that the determination result obtained by the second determination means is applied to the inspection target object which has become a non-defective product. Appearance inspection method. 4. The invention according to claim 1, wherein the first determination means mainly performs a determination process targeting a soldering state of the inspection object, and the second determination means mainly determines the inspection object. An appearance inspection method, characterized in that a determination process is performed for a mounting state of a shape of a component. 5. The invention according to claim 1, wherein the shape data pattern is a shape inclination angle code data distribution pattern.
Appearance inspection method characterized by being 6. A step illumination unit including a plurality of light sources, which are sequentially arranged above the inspection object in multiple stages and irradiate the inspection object with light at different angles, and the inspection by light irradiation from the step illumination unit. A plurality of imaging cameras that capture the reflected light from the surface of the object from directly above and diagonally above, an image memory unit that stores the image distribution data output from these imaging cameras, and for each pixel from this image distribution data An image calculation processing unit that performs calculation processing, a shape code data pattern unit that organizes a shape code data pattern from the calculation result of this image calculation processing unit, and a feature of the shape code data pattern is extracted to determine whether a good product or a defective product is extracted. By the recognition determination unit, the slit light generation unit disposed above the inspection target, and the slit light irradiated to the inspection target from the slit light generation unit An imaging camera that captures incident light from above with a certain angle as a light cutting line, an image processing unit that processes an image by the light cutting line from the output of this imaging camera to obtain shape line data, and this shape line data The position and shape of the inspection object is determined on the basis of the recognition determination unit that determines whether a good product or a defective product is extracted, and this recognition determination unit extracts a characteristic of the shape code data pattern from the good product and the defective product. It is configured to recognize and determine the ambiguous state of the inspection target by combining the extraction determination result and the determination result of the good product and the defective product obtained by obtaining the position and shape of the inspection target based on the shape line data. Appearance inspection device characterized in that 7. The invention according to claim 6, wherein an imaging camera that captures the reflected light from the surface of the inspection target due to the light irradiation from the step illuminating section from obliquely above, and the imaging camera disposed above the inspection target, An illuminating section for irradiating a plurality of lights sequentially arranged in multiple stages above the inspection object with different angles with respect to the inspection object, and reflected light from the surface of the inspection object by the light irradiation of the illuminating section. An imaging camera that captures from above and obliquely above, and an imaging camera that captures incident light from the slit light emitted from the slit light generation unit to the inspection object from above with a certain angle as a light cutting line. An appearance inspection apparatus, which is configured to be a common imaging camera. 8. The appearance inspection apparatus according to claim 6, wherein the step illuminating section is configured to perform illumination in a multi-step manner and in a multi-directional manner by sequentially switching in time.