JP2010038723A - Flaw inspecting method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flaw inspecting method which enables the accurate determination of quality by preventing the detection and over-detection of a flaw. <P>SOLUTION: The taken images inputted from a plurality of optical systems for regular reflected light and irregular reflected light are processed by an image processor and the flaw candidate extracted from the respective taken images by light and shade detection and flaw candidate feature quantity are subjected to composite processing to classify a flaw kind. Preset quality determination parameters different at each flaw kind are adapted to perform the proper determination of quality. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a defect inspection method for optically automatically detecting defects generated on the surface of a copper-clad laminate, for example, and classifying defect types and determining defect quality.

  A copper clad laminate such as a glass cloth epoxy copper clad laminate (CCL) has been strongly demanded for a product having a high appearance level. Accordingly, when an appearance defect is automatically detected, it is necessary to detect a defect signal with a low S / N ratio.

  In a conventional CCL defect inspection method, a captured image of a target object is obtained by a single or a plurality of illumination steps and a light receiving step, light and dark detection of the captured image is performed, defect candidates are extracted, and then a non-defective product and a defect are distinguished. A threshold is set, and a portion where the brightness of the defect candidate exceeds the corresponding value is regarded as a defect.

However, the conventional inspection method is effective only when it is assumed that the fatality of the defect is directly proportional to the size, depth, etc. of the defect. It is unsuitable for inspection objects where types of defects occur.
Further, in the high-resolution and high-precision inspection, there is a problem that the chance of extracting a formation other than a defect as a defect candidate increases, and “false information” increases.

  For this problem, as disclosed in the following Patent Documents 1, 2, and 3, it is more appropriate to classify a plurality of defect types into several categories and provide different defect pass / fail judgment criteria for each category. In addition, a method for determining whether or not a defect candidate is acceptable has been proposed.

  There are several types of copper-clad laminates with different pass / fail criteria such as, for example, that there are no dents with a size and diameter of 0.5 mm (φ) or more, and no scratches with a depth of 5 μm or more. There is a defect.

  Therefore, the defect pass / fail judgment criteria are not uniquely determined, and it is difficult to respond by setting the same pass / fail judgment threshold for all the conventional defect candidates, so the defect pass / fail judgment is made with higher accuracy. First, it is necessary to classify defect types and then determine whether or not the defect is good.

  In Patent Documents 1, 2 and 3, a defect inspection method for classifying a plurality of defect types having different defect quality determination into several categories has been proposed, but defects occurring on the product surface and defect quality determination criteria are It is an effective means only under certain conditions.

In addition, there is a method to classify the defect type with the shape of the defect area, aspect ratio, etc. of the captured image, but it is impossible to detect the whole defect part with a single optical system, and the whole defect is clarified Because it is not possible, the defect shape cannot be accurately analyzed.
Therefore, in order to classify defect types, it is necessary to make a judgment by combining the information detected across a plurality of optical systems.

Japanese Patent Laid-Open No. 04-066762 Japanese Patent Laid-Open No. 06-207909 Japanese Patent Laid-Open No. 06-029864

  It is an object of the present invention to provide a defect inspection method capable of preventing a defect from being detected and overdetecting and performing a quality determination accurately.

The present invention relates to a defect inspection method for inspecting a defect on a surface of a copper clad laminate, because of specularly reflected light on the surface of the inspection object in a direction parallel to or opposite to the movement direction of the inspection object and irregular reflection. An illumination step of irradiating light from each of a plurality of illumination sources for light;
The reflected light from the surface of the inspection object of each light irradiated from the plurality of illumination sources to the surface of the inspection object by the illumination step, the direction orthogonal to the movement of the inspection object or the direction opposite to the movement direction A light receiving step for receiving light by a plurality of light receiving means having a light receiving operation direction;
An image in which defect candidates are extracted by performing image processing on each of the plurality of captured images obtained by the light receiving step and detecting feature regions of the inspection object by light and dark detection, and performing analysis of a plurality of feature amounts of the feature regions Processing steps;
A combined processing defect classification step for associating image processing results of a plurality of captured images of the image processing step, performing composite processing, and classifying the types of defects of the inspection object;
Defect type pass / fail determination step for determining pass / fail according to different defect pass / fail determination conditions for each classified defect candidate,
It is related with the defect inspection method characterized by comprising the process containing these.

  Further, according to the present invention, the composite processing defect classification step performs binarization processing on each of the plurality of captured images obtained by the light receiving step in the image processing step, and obtains binarized data by performing binarization processing. Then, defect candidates are extracted by a labeling process (pixel concatenation process), and feature quantities are obtained from a plurality of particle analyzes of aspect ratio, circularity, and perimeter for the defect candidates, and extracted from a plurality of captured images. The present invention relates to the defect inspection method described above, wherein the defect candidate and the feature amount of the defect candidate are associated with each other based on the information on the occurrence position of the defect candidate, and the feature amount of the defect is processed in a composite manner.

  Further, according to the present invention, the combined processing defect classification step synthesizes using a plurality of defect feature quantities combined by associating the defect candidates and the feature quantities of the defect candidates, and performs conditional branching to determine the defect type of the defect candidate. The present invention relates to the above-described defect inspection method characterized by performing sorting.

  Further, according to the present invention, in the defect type pass / fail judgment step, for each defect type group classified by the combined processing defect classification step, defect quality judgment criteria of area, brightness and darkness level are individually set for each defect type. The present invention relates to the defect inspection method described above.

  According to the present invention, conventionally, defect determination is performed for all detected defect candidates by determining the same pass / fail judgment criteria without performing defect type discrimination, whereas defects are detected for detected defect candidates. It is possible to perform defect classification processing and perform more appropriate pass / fail judgment for a plurality of types of defects each having different pass / fail judgment criteria, thereby preventing defect non-detection and overdetection.

The best mode for carrying out the invention will be described below with reference to the drawings.
FIG. 1 is a system configuration diagram of a defect inspection apparatus for explaining the present invention.
This defect inspection apparatus is an apparatus for inspecting defects on the surface of a copper clad laminate such as a glass cloth epoxy copper clad laminate (CCL) as an object to be inspected, and executes the defect inspection method of the present invention.

The defects of the copper clad laminate are roughly classified into a concave defect, a scratch defect and a smooth defect.
If these defects exceed a certain standard, disconnection may occur during circuit processing, causing problems in product function, and the standards causing problems in product function are different for each defect. .

Concave defects include dents, copper troughs, cracks, and troughs.
The dent is generated when a foreign substance enters between a copper foil and a jig plate required for press forming during press forming.
The copper cocoon is a copper foil cocoon, and is caused by the thin copper foil.

The crack is a crack in the copper foil at the resin protrusion (tread) portion of the prepreg, and is caused by the severeness of the copper foil due to the crack.
Also, the cocoon cake is a copper cocoon at the heel part, and is caused by the mild twist of the copper foil by the cocoon.
For example, the criticality of these increases depending on the size of the defect, such as a size and a diameter of 0.5 mm (φ) or more.

  In addition, the flaw defect is a cause of troubles at the time of transportation by a manufacturing facility. The fatality increases in proportion to.

In addition, smooth defects include discoloration and deposits. The discoloration is rust and dirt on the copper clad laminate copper foil, and is caused by water residue on the end plate.
Further, the deposit is caused by foreign matters generated on the surface of the copper-clad laminate, and is caused by adhesion of machine oil or the like. These are fatal in proportion to the size of the defect.

It is necessary to detect these multiple defects stably and make an appropriate pass / fail judgment.
First, in order to obtain a high detection capability, the illumination / imaging system has at least two optical systems of specular reflection illumination / imaging system that emphasizes smooth defects such as color unevenness and irregular reflection illumination / imaging system that emphasizes irregular defects. In addition, in order to perform appropriate pass / fail determination, defect type classification processing is performed by performing combined processing of regular reflection illumination / imaging system and irregular reflection illumination / imaging system.

  The defect inspection apparatus in FIG. 1 irradiates light onto a copper-clad laminate, which is an object to be inspected, placed on a conveyor, and captures an image and an image sent from the illumination / imaging system. An image processing apparatus that performs image processing, which will be described later, and a composite processing apparatus that performs composite processing on the image processed image to classify defect types and determine pass / fail are provided.

  The illumination / imaging system irradiates two or more illumination sources that irradiate with parallel rays from the opposite or moving direction of the inspection object, the two or more illumination sources, and the surface of the inspection object. In addition, each light is composed of two or more light receiving means having a light receiving operation direction in a direction orthogonal to the direction opposite to the movement of the inspection object or the moving direction.

The image processing apparatus captures an image in the light receiving step, captures an image of the inspection object, and then performs image quality improvement processing and binarization / labeling processing to extract defect candidates.
Further, the shape analysis of the extracted defect candidates such as the aspect ratio and the perimeter is performed.

  Further, the composite processing device associates the shape analysis results of defect candidates extracted by a plurality of image processing devices such as a regular reflection image processing device and a diffuse reflection image processing device based on the defect occurrence position, and Integration of the defect feature values made is performed.

FIG. 2 is a functional block diagram of the image processing apparatus in the defect inspection apparatus.
The image processing apparatus functionally includes a shading correction processing unit, a noise filter unit, a binarization processing unit, a labeling unit, and a feature amount analysis unit.

The shading processing unit performs shading correction processing on the specular reflection image, irregular reflection image, and the like captured at the light receiving step to correct the received light amount variation.
The noise filter unit filters noise generated by an image sensor, a change in conveyance speed, and the like by an image filter on the captured image that has passed through the shading correction processing unit.

The binarization processing unit binarizes the captured image that has passed through the shading correction processing unit and the noise filter unit.
The labeling unit extracts defect candidates by obtaining a connected component of the characteristic particle group obtained by the binarization process through the shading correction processing unit, the noise filter unit, and the binarization processing unit.

The feature amount analysis unit performs a defect such as the number of connected pixels, a short diameter, a long diameter, etc. on the defect candidate group extracted through the shading correction processing unit, the noise filter unit, the binarization processing unit, and the labeling processing unit. The feature analysis is performed.
All the obtained defect candidates and defect feature quantities of each optical system are transferred to the composite processing apparatus through the feature quantity analysis unit.

FIG. 3 is a functional block diagram of the composite processing apparatus in the defect inspection apparatus.
The composite processing apparatus functionally includes a defect candidate association processing unit, a defect candidate feature amount integration processing unit, a defect type classification processing unit, and a defect quality determination processing unit.

  The defect candidate associating processing unit calculates the defect candidate and the defect feature amount in the captured image transferred from the image processing apparatus of each optical system based on the defect candidate extraction position in the periphery of the plurality of other optical system captured images. A search is performed, and those having substantially the same coordinates are regarded as the same defect and are associated.

  The defect candidate feature amount integration processing unit performs defect feature amount integration such as adding the areas to the defect candidate group subjected to the association processing through the defect candidate association processing.

  The defect type classification processing unit passes through the defect candidate association processing unit and the defect candidate feature amount integration processing unit, based on the defect feature amount possessing defect shape parameters and the like of the defect candidate group for the combined defect candidate group. In addition, defect types are classified by conditional branching.

  The defect pass / fail determination processing unit performs pass / fail determination for each defect type group with respect to the defect candidates subjected to the defect type classification through the defect candidate association processing unit, the defect candidate feature amount integration processing unit, and the defect type classification process. The defect candidate pass / fail judgment process is performed by applying defect pass / fail judgment parameters with different standards.

FIG. 4 shows a defect candidate associating process technique of the composite processing apparatus of the defect inspection apparatus. That is, the processing procedure of the composite processing of the present invention executed by the composite processing apparatus is shown.
The two images are images in which the same scratch defect is captured by two different optical systems A and B.
First, when defect candidates are extracted by the optical system A, the defect occurrence position coordinates of the defect candidates are calculated.

  Next, in the other optical system B, a peripheral search of the occurrence position coordinates of the defect candidate extracted in the optical system A is performed, and when the defect candidate extracted in the other optical system B is found, the optical system A extracts it. The associated defect candidate and the defect candidate extracted by the other optical system B are associated with each other.

FIG. 5 is an example of a flowchart of the defect type classification process of the composite processing apparatus in the defect inspection apparatus.
For example, conditional branching as shown in FIG. 5 is performed based on the feature amount of the defect candidate associated with the defect candidate associating process.

  As described above, according to the defect inspection method of the present invention, defects input from a plurality of optical systems in detection of a plurality of types of defects having different quality determination criteria and defect quality determination that occur on the surface of a copper-clad laminate. By combining candidate signals, defect types can be classified with higher accuracy, and different pass / fail judgment parameters can be set for defect groups with different pass / fail judgment criteria, thus preventing defect non-detection and over-detection. can do.

It is a system configuration figure of a defect inspection device for explaining the present invention. It is a functional block diagram of the image processing apparatus in a defect inspection apparatus. It is a functional block diagram of the composite processing apparatus in a defect inspection apparatus. This is a defect candidate associating process method of the composite processing apparatus in the defect inspection apparatus. It is an example of the flowchart of the defect kind classification | category process of the compound processing apparatus in a defect inspection apparatus.

Claims (4)

In a defect inspection method for inspecting a defect on a surface of a copper clad laminate, for a specular reflection light and a diffuse reflection light on the surface of the inspection object with a parallel light beam from a direction opposite to or in a movement direction of the inspection object An illumination step of irradiating light from each of a plurality of illumination sources;
The reflected light from the surface of the inspection object of each light irradiated from the plurality of illumination sources to the surface of the inspection object by the illumination step, the direction orthogonal to the movement of the inspection object or the direction opposite to the movement direction A light receiving step for receiving light by a plurality of light receiving means having a light receiving operation direction;
An image in which defect candidates are extracted by performing image processing on each of the plurality of captured images obtained by the light receiving step and detecting feature regions of the inspection object by light and dark detection, and performing analysis of a plurality of feature amounts of the feature regions Processing steps;
A combined processing defect classification step for associating image processing results of a plurality of captured images of the image processing step, performing composite processing, and classifying the types of defects of the inspection object;
Defect type pass / fail determination step for determining pass / fail according to different defect pass / fail determination conditions for each classified defect candidate,
A defect inspection method comprising the steps including:   In the composite processing defect classification step, after the image quality improvement processing is performed on each of the plurality of captured images obtained by the light receiving step in the image processing step, binarization processing is performed to obtain binarized data, and labeling processing (pixel The defect candidate is extracted by a concatenation process), and the feature amount is obtained by analyzing a plurality of particles of aspect ratio, circularity, and perimeter for the defect candidate, and the defect candidate and defect candidate extracted from the plurality of captured images are acquired. 2. The defect inspection method according to claim 1, wherein the feature quantity is correlated based on the information on the occurrence position of the defect candidate, and the feature quantity of the defect is processed in a complex manner.   The combined processing defect classification step uses a plurality of defect feature values combined by associating the defect candidates and feature values of defect candidates, and performs defect branching of defect candidates by conditional branching. The defect inspection method according to claim 1, wherein:   The defect type pass / fail judgment step individually determines and applies defect quality judgment criteria of area, brightness and darkness level for each defect type for each defect type group classified by the combined processing defect classification step. The defect inspection method according to claim 1, wherein:

Images