JPS6026175B2 - Defect detection method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a defect detection method, and more particularly to a method for automatically detecting various defects that would be disqualified if present on the surface of a three-dimensional object or a sheet-like object.

Most of the defects present on the surface of such objects often have an optically different appearance compared to the rest of the object.

For example, stains on yarn rolls such as burns or on the surface of cloth,
The superior part can be recognized as having a brightness that is clearly different from other parts. Also, defects that are difficult to see under normal lighting (
For example, by changing the direction of illumination, uneven parts can be easily recognized as having a brightness different from other parts. Conventionally, as a method for detecting such defects, for example, as an inspection method for sheet-like materials, there was
127279, a method for inspecting thread scrolls is, for example, a method based on Jitsoseki 49-112951, and a method for inspecting metal plates is, for example, 49-112951 Tsubaki.
A method typified by No. 6-21515 has been proposed.

The device proposed by the first Japanese Patent Publication No. 43-27279 has a number of photoelectric elements arranged along the width direction of the sheet to be sensitive to the transmitted or reflected light irradiated onto the sheet being transported in the length direction. They are arranged side by side and the optical turtle elements are vibrated in the width direction of the sheet.The device proposed by the second author, Seki Sho 49-112951, applies direct current or high frequency current to the surface while rotating the winding yarn. Light is emitted by a discharge lamp, and changes in the amount of reflected light are received by a light-receiving element such as a solar cell, which is converted into a current change, and a current change greater than a reference value is detected. These two methods have drawbacks such as low resolution, inability to detect minute defects, and limitations on the size and shape of the object to be inspected because many single elements are arranged one by one as light receiving elements. . In addition, the device proposed by the third Tsubaki Sekisho 46-21515 reproduces a complete electrical signal of one field captured by an electro-optical device that scans an image of an object with an electron beam to eliminate signals caused by surface defects. In order to extract the pulse, a level analysis is performed and the pulse is integrated to evaluate the optical appearance of the object. This device had problems in that it could not be used to inspect objects with unique shapes, such as the foot of a barn, and was subject to limitations in terms of use. The purpose of the present invention is to provide a versatile method for automatically detecting such defects that can be handled optically, and more specifically, to provide a solid-state device with high performance as a detection sensor. Another object of the present invention is to provide a method for detecting defects by scanning an object to be inspected using a solid-state camera capable of selecting a scanning speed over a relatively wide range and capable of high-speed scanning and processing the output signal.

In order to achieve this object, the defect detection method according to the present invention scans an object to be inspected with a solid-state camera in which light-receiving elements are linearly arranged, and detects defects existing in the object based on the output signal of the solid-state camera. The method includes: a mechanism for sequentially comparing the aberration value of the output signal between the light receiving elements in each scan with a reference value of 1; A mechanism that calculates the column difference between the value at the time of scanning before and after that and sequentially compares the column difference with the second reference value, and a mechanism that adds the column difference of the output signal for each sampling by a fixed number for each of the light receiving elements. The present invention is characterized in that a moving average is calculated using a moving average value and a mechanism that compares each moving average value with a third reference value is used to separate the signal due to the defect.

Embodiments of the defect detection method according to the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of the present invention, in which the output of the solid-state camera 1 is supplied to the hold amplifier 3.
/D converter 4 converts it into a digital signal.

Next, information for one line of the digital signal is temporarily stored in a memory 5 having a capacity corresponding to the number of conversion bits times the number of elements.
In the figure, the solid-state camera is controlled by a camera control unit 2, and a memory control circuit 15 receives a scanning pulse from the camera control unit 2 and generates a start pulse, a stop pulse, and a timing pulse in order to write information into the memory 5. to control the memory 5. When the writing of the information corresponding to one scan of each element converted into a digital signal into the memory is completed, the information is supplied to the sequential arithmetic circuit 6 in accordance with the input order. The circuit 6 is a circuit that calculates the front and back output difference (line difference) between elements, and here, if each element output is Dii~D(j+n)i, ID(i
+1) i-Diil, ID (j+2) i-D (j+1)
il... is calculated. The results of each calculation are then supplied to the digital comparator circuit 8. The comparison circuit 8 receives each input information D. is compared with a reference value ds, specified in advance by the setting device 9, and the result is ds. ≦D
, an abnormal signal A is output under the conditions of . Therefore, here, elongated defects in the direction perpendicular to the element arrangement direction can be separated more sharply. Next, the arithmetic circuit 7 is a circuit that calculates the difference between before and after outputs (column difference) for each element at every sampling, and the arithmetic circuit 7 receives signals from the memory 5 and the A/D
When the signal from the converter 4 is being supplied and the acquisition into the memory 5 is completed, the information is sequentially read out from the memory 5 in synchronization with the timing of the output signal from each element in the next sampling, and IDi(i+1-Diil , ID(i
+1)(i+1)-D(j+1),...l(D(j+n
)(i+1)-D(j+n)i l, IDj(i+2)
−Dj(i+1)l, ID(j+1)(i+2)−D(
j+1)(i+1)l...ID(j+n)(j+2)-
D(j+n)(i+1)l...ID(i+n)(i+n
)−D(i+n−1)(i+n−1)l,
The result is supplied to the digital comparator circuit 10.

Here, the calculation result D2 of each element output supplied from the calculation circuit 7 is compared with the reference value ship 2 by the reference value setter 11, and when the result is ds2-D2, an abnormality signal A2 is output. In other words, defects that are long in the element arrangement direction b are also sharply separated. On the other hand, the output from the arithmetic circuit 7 is also supplied to the arithmetic circuit 12, and the circuit 12 is a circuit that calculates a moving average in the element arrangement direction for a plurality of arithmetic result data of the arithmetic circuit 7. For example, Dj(i+1
)−Dii=△D(i・j), then l! History△
Calculate D(i·i)l=D3. Next, this 1=J The result is supplied to the digital comparator circuit 13 and the reference value a
It is compared with s3 and when the result is as3≦D3, an abnormal signal is output.

In other words, this process separates defects that have a relatively large area and a small level change. As described above, the present invention has three signal discrimination functions, and the solid-state camera output signal for each sampling is processed by each of these discrimination mechanisms to separate abnormal signals.

By this process, defects that reflect a larger amount of light than normal parts or small defects can be reliably separated regardless of the size of the area. Next, to explain an example, FIG. 2 is a diagram showing an example of the inspection of a sheet-like material 16 such as paper, film, cloth, etc., in which the roller 17,
A line scan type solid-state camera is mounted above the traveling sheet-like object 16 held between 17' and 18, 18', and an optical system is installed so that the element arrangement direction corresponds to the width direction of the sheet-like object 16 and covers the entire area. Adjust and place.

Here, when light is irradiated from slightly above the sheet-like object 1, defects such as a2 protruding from the sheet-like object will have a positive shadow, and a waveform as shown in the waveform opening will be obtained as an output, and other defects a3 will have a waveform. A waveform as shown in (a) is obtained as an output, and for a defect (a) etc. that is elongated in the element arrangement direction, a waveform (a) as shown in waveform (a) is obtained as an element output waveform for each sample. Therefore, in the discrimination circuit, among these, defect a is A2, and defect a2 is A2.
, the output defect a3 is separated as an output. Furthermore, by outputting the defect signal thus separated, a control mechanism as shown in Fig. 4 controls a relay circuit through a logic circuit OR, for example, to stop the rotation of the motor, indicate an abnormality with a lamp, or move to an abnormal location. It becomes possible to carry out markings, etc. Next, Fig. 3 shows a practical example of the inspection of a thread roll as an application to the inspection of three-dimensional objects. arranged so that the element arrangement direction corresponds to the longitudinal direction of the barn 19,
By rotating the barn, defect division 4 (stain defect), minor (uneven defect), and common (thread stain defect) are sequentially changed to waveform A,
By the processing mechanism of the present invention, the defect A4 is obtained as the analog output shown in Figs. can be separated as output.

Accordingly, the output enables various actions such as defect display marking by a control mechanism as shown in FIG. 4, for example. Note that it is also possible to apply the present invention to the inspection process for other three-dimensional objects such as ampoules of black ink. Above, we have described only the case where a line scan type solid-state camera is used as a practical example of the present invention, but most of these inspection processes involve moving in a fixed direction at a constant speed, and the line scan type solid-state camera is Number of elements (resolution) of the camera
This is because it is very convenient, and if the speed, resolution in the line direction, etc. are satisfactory, there is no problem with an area type.

As described above, the present invention not only enables the automation of various inspection processes, but also enables the quantification of quality control when connected to a computer, and its effects are extremely large.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a configuration example of the defect detection method according to the present invention, FIGS. 2 and 3 are examples of the defect detection method according to the present invention, and FIG. 4 is a control system of the method according to the present invention. This is an example. In the figure, 1 is a solid-state camera, 2 is a camera control unit, 3 is a hold amplifier, 4 is an A/D converter,
5 is a memory, 6, 7 and 12 are arithmetic circuits, 8, 10,
13 is a digital comparator; 9, 11 and 14 are first, second and third reference value setters, respectively; and 15 is a memory control circuit. Figure 2 Figure 3 Figure 4

Claims (1)

[Claims] 1. In a method of scanning an object to be inspected with a solid-state camera in which light-receiving elements are linearly arranged, and inspecting defects existing in the object using the output signal of the solid-state camera, the output signal is transmitted to the A mechanism that sequentially compares the row difference value of the output signal between the light receiving elements with a reference value of 1, and a column difference between the output signal of each of the light receiving elements and the value at the time of scanning before and after each sampling of the output signal. and a mechanism for sequentially comparing the column differences with the second reference value; 3. A defect detection method characterized in that the signal due to the defect is separated by processing using a mechanism for comparing with the reference value of No. 3.