JP2002031605A - Defect-confirming apparatus and automatic visual inspection apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that inspected results by a visual inspection apparatus not only include false reports because the apparatus processes images by special hardware and is unable to give a detailed judgment, but are not correctly sorted and cannot be practically used as quality management data because defects of pattern similarities even if detected by the apparatus are sometimes regarded as non defects at a confirmation work. SOLUTION: A defect-confirming apparatus is provided with an image processor to execute more detailed inspection during the confirmation work. Non defects are eliminated from confirmation objects by a worker.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic inspection apparatus for wiring patterns, and more particularly to a confirmation apparatus attached to an automatic appearance inspection apparatus for wiring patterns.

[0002]

2. Description of the Related Art A commercially available automatic visual inspection apparatus for a wiring board generally includes an illumination device for illuminating a surface to be inspected, and a CC.
A method of imaging a wiring pattern by a camera such as a D linear sensor or a method of scanning with a laser beam and detecting the wiring pattern with a photomultiplier tube or the like to image the wiring pattern is adopted. Since these inspection apparatuses need to end the inspection within the real time of scanning the camera, image processing and defect determination are performed by hardware having a dedicated pipeline configuration. These dedicated hardware apply the imaged pattern to the pre-adopted template and extract the part that matches the defect feature, or check the consistency between the extracted feature and the feature extracted from the non-defective pattern. In general, a method of detecting a defect in a pattern by examining the pattern is used. One of the drawbacks of image processing using such dedicated hardware is that patterns with certain characteristics can easily be pointed out as false information even if they are not defects. In reality, there is a difference between the displayed defect classification and the shape of the actual defect by hand. Therefore, even if an attempt is made to use the result indicated by the inspection apparatus for the data of quality control, it is not practical.

[0003] The defect checking device compensates for the above-mentioned drawbacks of the inspection device.
By reconfirming by the operator, false alarms and actual defects are distinguished. The structure of the defect confirmation device is an XY stage and a TV
It consists of a camera, TV monitor, and lighting, and has no special image processing unit because its function is only to enlarge and display the pointed part.

[0004] However, when it is desired to accurately classify the defects, the operator must classify them. However, even if the defects are recorded in a work ledger or a defect type is inputted from a keyboard, the efficiency of the inspection work is reduced. Therefore, the fact is that it is not generally performed. In addition, when there are many false reports in the inspection apparatus, it is necessary to confirm all the defects, so that the efficiency may be extremely deteriorated. For example, 200 per side
In the event of false alarms, workers must visually check all 200 locations and identify them as true defects, which can be very painful and can be overlooked by workers. The sex is great. Assuming that the confirmation time per case is about 1.5 seconds, the number of workers is 200 cases / surface × 1.
It takes 5 seconds / case = 300 seconds / surface. This means that the inspection time of most currently available inspection equipment is 6 times per surface.
It is fatally slow considering that it is 0 seconds or less.

[0005] Further, since the defect confirmation device only visually confirms the indicated defect, a defect similar to the design pattern, loss of the pattern, and a solid short-circuit of the power supply and the ground pattern are detected by the inspection device. In such a case, there is a problem that the worker cannot be recognized as a defect and is overlooked.

Some automatic appearance inspection apparatuses have a defect confirmation function, but the function has a similar defect.

[0007]

An object of the present invention is to make it possible to classify defects which do not reduce the throughput and which can be used for quality control data even when there are many false reports from the inspection apparatus. Another object of the present invention is to provide a defect checking device that is hardly overlooked by an operator.

[0008]

As described above, the image processing apparatus of the visual inspection apparatus is dedicated hardware, but a personal computer (PC) or a workstation (WS) is used as the image processing apparatus. Image processing can also be performed by software. Compared to the above-mentioned dedicated hardware, it is possible to perform various image processing such as length measurement of defect parts, comparison with product specifications, defect classification, etc. As a result, false alarms can be greatly reduced and defect classification is easy However, the image processing speed per frame is reduced, and the inspection speed is determined by the image processing apparatus. For example, considering a device that captures a field of view of about 5 mm using a camera of 512 pixels × 512 pixels, 500 mm × 600 mm
In order to inspect the entire surface of the printed circuit board, it is necessary to inspect 12000 frames. Even if the inspection time required for one frame is 0.3 seconds, it takes 3600 seconds, that is, one hour. For the reasons described above, PCs and WSs are not suitable as image processing devices for visual inspection devices used for production.

However, by mounting such an image processing apparatus on a defect confirmation apparatus and re-inspection of a point indicated by the inspection apparatus by the defect confirmation apparatus again, the above-mentioned problem of the defect confirmation apparatus can be solved. The time required for a worker to visually recognize and judge one pointed point is more than one second. If the defect can be determined by performing image processing within this time, the inspection can be performed without giving any stress to the worker. Since the speed of image processing is at most about 0.3 seconds per frame even if the image processing is performed by software, the determination is completed before the operator visually checks the image. However, even in this case, the occurrence of a false report is inevitable, so that the worker needs to confirm the work, but it is sufficient to visually confirm only when the defect determination apparatus again indicates the defect. Although an image processing apparatus using dedicated hardware emphasizing speed is expensive, a low-cost PC-based image processing apparatus can be sufficiently mounted on a confirmation device.

In the case of image processing by software, it is not difficult to classify defect shapes with high accuracy. In addition, by connecting to CAD, the checking operation can be performed while referring to the design pattern. Therefore, even if the defect is easily overlooked by the operator, for example, a defect similar to the design pattern, a pattern loss, a solid short of a power supply or a ground pattern, or the like. It can be confirmed.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be specifically described below.

FIG. 1 is a diagram showing an outline of the present invention.
Reference numeral 1 denotes a defect confirmation device according to the present invention. A printed board 3 inspected by a visual inspection device 3 is fixed to a confirmation stage 2 with guide pins 4. Appearance inspection device 16
The coordinate data of the defect detected by the visual inspection device 16 through the Ethernet (registered trademark) 14 is transmitted from the PC 11.
Sent to First, the defect confirmation device 1 is fixed to the XY stage 5 by a command from the keyboard 10.
By driving the 2 × 512 pixel TV camera 7,
Move up. The defect portion 8 is illuminated by the illumination device 6, and an image is captured through the TV camera 7. An image from the TV camera 7 is captured through an image input board in the PC 11. A design pattern generator 17 is connected to the Ethernet 14, generates a local design pattern corresponding to the pointed point 8, and transmits it to the PC 11. PC1
Reference numeral 1 denotes an image processing program for performing image processing on a captured image to perform defect determination and defect classification. The classified result is displayed on the operation screen 12, adds a defect code to the coordinate data, and is transferred to the quality management database 18. The TV camera 7 is programmed to automatically move to the next defect if the captured pattern is determined to be non-defective, and to wait for keyboard input if determined to be defective, and to be programmed not to move to the next defect. ing. The worker is a TV monitor 9
In addition to being able to observe the enlarged image of the pointed portion 8, it is also possible to refer to the design pattern on the TV monitor 9. If there is a problem in the classification of the defect, or if it is a false report, the defect classification can be corrected from the keyboard here. When the enter key is pressed, the camera 7 moves to the next defect.

FIG. 2 shows an example of text data of data transferred to the defect confirmation device. () Indicates the description of the data,; indicates a line feed code, and // indicates the end of the file.
Data includes product name, serial number, date, origin offset XY coordinates, total number of indications, number of false reports, number of disconnections, number of missing parts,
The number of shorts, the number of protrusions, and the number of remaining isolated coppers are at the head, and thereafter, a defect number, a defect XY coordinate, and a defect code are recorded for the indicated number. As for the defect code, 0 is unconfirmed, 1 is false information, 2 is broken, 3 is missing, 4 is short, 5 is protrusion, 6
Indicates an isolated copper residue. In this example, there are 15 indications, and the confirmation up to the fourth has been completed. The defect code is automatically input, but if there is a determination error, it can be corrected by an operator. When all operations are completed, this data is transferred to the quality management database.

[0014]

As described above, according to the present invention, the location detected by the inspection device is automatically inspected again in detail by the confirmation device, so that false information of the inspection device can be eliminated and the design pattern can be referred to. By doing so, the burden on the worker can be reduced, and oversight by the worker can be reduced. In addition, since the classification of defects is not the level pointed out by the inspection equipment, false alarms are excluded and classified correctly, using a database to display the number of defects generated for each defect,
It is possible to search in chronological order and to display a map of a location where a disconnection has occurred, thereby contributing to quality control.

[Brief description of the drawings]

FIG. 1 is a schematic diagram of an embodiment of the present invention.

FIG. 2 is a diagram illustrating an example of text data of data transferred to a defect confirmation device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Defect confirmation device, 2 ... Confirmation stage, 3 ... Printed circuit board, 4 ... Guide pin, 5 ... XY stage, 6 ... Illumination device, 7 ... TV camera, 8 ... Pointed-out point, 9 ... TV monitor,
10 keyboard, 11 PC, 12 operation screen, 13
... XY axis motor controller, 14 ... Ethernet, 1
5: transceiver, 16: visual inspection device, 17: quality control database.

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 2F065 AA49 BB02 CC01 DD06 FF04 JJ19 MM24 QQ25 QQ31 RR05 2G051 AA65 AB02 AC01 AC15 CA04 CB01 EA14 EA21 EC01 ED11 FA04 5B057 AA03 BA02 DA03

Claims (5)

[Claims] 1. A checking and correcting device for an operator to visually check or correct a location pointed out by a wiring pattern defect inspecting device as a defect, comprising: a camera for detecting an image; and an image for processing a captured image. Equipped with a processing device, characterized in that the point pointed out by the inspection device is inspected again in detail,
Defect confirmation device. 2. An inspection apparatus having a defect confirmation function, wherein a point pointed out by the inspection apparatus is inspected in detail again by an image processing apparatus different from that at the end of inspection or at the time of defect confirmation or inspection. , Automatic visual inspection equipment. 3. An image processing apparatus having a plurality of image processing apparatuses, inspecting the entire surface with a first image processing apparatus, and re-detailing a portion pointed out by a (n-1) th processing apparatus with an nth image processing apparatus. Automatic appearance inspection device characterized by performing inspections. 4. A design pattern generating device, wherein by referring to a design pattern of the defective portion, it is possible to determine a defect such as a missing pattern which is difficult for an operator to identify as a defect. Defect checking device or automatic visual inspection device. 5. The defect checking device or the automatic visual inspection device according to claim 1, wherein the image processing device has a function of automatically classifying the type of the defect pointed out by the defect inspection device as a defect.