JPH1048151A - Surface flaw detection method - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To provide a method capable of separating and detecting individual flaws and correctly detecting and determining various forms of flaws over a plurality of screens. SOLUTION: In the method of irradiating the surface of a band-shaped body with light and capturing the reflected light with a plurality of light receivers arranged in the width direction to detect surface flaws, candidates of flaws detected in a certain visual field A method of detecting a surface flaw based on information of a flaw in a plurality of fields of view, by adding information indicating a connection state at a boundary with a field of view of an adjacent light receiver together with a feature amount of the flaw within the field of view. Similarly, in a method of inspecting a surface flaw by dividing a screen in the longitudinal direction, a candidate for a flaw detected in a certain screen, along with the feature amount of the flaw in the screen, a screen adjacent to the width direction and the longitudinal direction. A surface flaw detection method for detecting surface flaws based on information on flaws on a plurality of screens by adding information indicating a connection state at boundaries a to h with respect to the surface.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting surface flaws of a belt-like body and a general method of detecting surface flaws, including detection of surface unevenness such as an inspection of a printed state.

[0002]

2. Description of the Related Art Generally, a method of optically detecting a surface flaw of a belt-like body such as a thin steel plate or a resin film is used.

FIG. 5 is an external view of a typical apparatus used in this method. A laser beam is scanned by a polygon mirror, information on the surface to be inspected is captured by a light receiver, and a flaw is detected by a signal processing device.

In general, the detection of a flaw is performed by dividing information on a surface to be inspected into a screen of a predetermined size (the number of pixels) due to restrictions on the apparatus. At this time, a method of detecting a flaw extending over a plurality of screens and a method of determining the flaw are important issues. In the technique described in Japanese Patent Application Laid-Open No. 4-110758, image processing of flaws is performed by a profile method, and the characteristics of the flaws are arithmetically processed from the results.

[0005] The profile method described in this publication is
This is a method of projecting an image of a flaw in the longitudinal direction and converting it into a distribution form in the width direction. The type of the flaw is determined from the information on the distribution form in the width direction, and if the distribution of a flaw in the width direction extends over a plurality of visual fields, it is determined that the flaw is continuous.

[0006]

Problems to be Solved by the Invention
In the technique described in Japanese Patent Application Laid-Open No. 8 (1999) -2000, a distribution form obtained by projecting a flaw image in the longitudinal direction is used as data for determination, and thus an erroneous determination may be made. For example, even when there are actually a plurality of flaws in the visual field, when the flaw image is projected in a certain direction, these flaws often appear to overlap. In this case, these flaws are determined as one continuous flaw in the width direction as a distribution form, and the number and size of the flaws are detected differently from actual ones.

Information obtained on each screen (distribution form)
A relatively large storage capacity is required to save the data even after the determination of the flaw. Therefore, it is difficult to preserve the distribution form even after the determination of the flaw, and the flaw information cannot be confirmed again later.

The present invention provides a method for separating and detecting individual flaws and for correctly detecting and determining various forms of flaws over a plurality of screens.

[0009]

According to a first aspect of the present invention, there is provided a method for detecting surface flaws by irradiating light to a surface of a strip and capturing the reflected light with a plurality of light receivers arranged in a width direction. For a flaw candidate detected in a certain field of view, together with the feature amount of the flaw in the field of view, information indicating the connection status at the boundary with the field of view of an adjacent light receiver is added,
A surface flaw detection method characterized by detecting surface flaws based on information on flaws in a plurality of visual fields.

Here, the information indicating the connection status at the boundary with the field of view of the adjacent light receiver includes information as to whether or not a part of the flaw is at the boundary with the field of view of the adjacent light receiver.
If there is, it is used to mean information on the position at the boundary.

If no detected flaw candidate is on the boundary,
That is, when the flaw is within the field of view, the flaw candidate is determined as a flaw represented by the feature amount. Here, the characteristic amount of the flaw can be appropriately selected as needed, as to the size, area, hue, shape, and the like of the flaw.

If the detected flaw candidate is on the boundary, there is a possibility that the flaw candidate is continuously present in the adjacent visual field. Therefore, for a flaw candidate in an adjacent visual field, an object (flaw candidate) that is continuous with the flaw candidate in the original visual field is selected using information indicating the connection status at the boundary with the original visual field. If there is only one flaw candidate, it is continuous with the flaw candidate in the original field of view. When there are a plurality of flaw candidates on the corresponding boundary, a correspondence with the flaw candidate in the original visual field is further made by using information on the position at the boundary.

In this way, when the information indicating the connection status of the flaw candidate to the adjacent visual field is interrupted, the characteristic amount of the flaw in each visual field where the flaw exists is added to determine the flaw. The summation of the flaw feature amounts generally uses a total value for length and area, an average value for hue and the like, and a necessary method for shape and the like.

According to a second aspect of the present invention, the surface of the strip is irradiated with light, the reflected light is captured by a plurality of light receivers arranged in the width direction, and the surface is inspected by dividing the screen in the longitudinal direction. In the method, for a flaw candidate detected in a certain screen, information indicating a connection state at a boundary between screens adjacent in the width direction and the longitudinal direction is added together with the feature amount of the flaw in the screen, and A surface flaw is detected based on the flaw information on the screen.

The present invention is a two-dimensional extension of the first aspect of the present invention, wherein information on whether or not a flaw is continuous with an adjacent area is also provided in the longitudinal direction, that is, on the boundary of the area which is temporally successive. Add. In this way, even if the processing of the flaw detection is divided in the longitudinal direction, the feature amount of the whole flaw continuous in the longitudinal direction is obtained, and the surface flaw can be detected with high accuracy.

Further, according to the present invention, from the information indicating the connection state of the flaws at the overlapping part of the boundary in the width direction and the boundary in the longitudinal direction, that is, at the four corners of the screen, the flaws are continuously connected to the obliquely adjacent area. It is also possible to determine whether or not there is.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a diagram showing a certain visual field and the form of a flaw at its boundary. In the figure, the solid rectangle indicates the entire visual field range, and the portion outside the dashed line indicates the boundary area in the full visual field. This boundary area is an area for collecting information indicating whether or not the flaw is continuous with an adjacent visual field, that is, information indicating the connection state of the flaw. Here, the boundary region of the rectangular viewing range is divided into eight regions a to h corresponding to the four sides and the four corners of the rectangle. The region i is an internal region.

Areas d and h indicate the connection with the field of view (adjacent in the width direction) of the adjacent photodetector, and areas b and f indicate the state of connection with the field of view that is temporally back and forth (adjacent in the longitudinal direction). This is an area for collecting each. Four corner areas a, c,
“e” and “g” are regions for collecting the connection status with the oblique view when the longitudinal direction and the width direction are the vertical and horizontal directions, respectively.

When a signal of a flaw candidate is detected in these boundary areas, information indicating that there is a connection with an adjacent visual field is added as information indicating a flaw connection state together with the feature amount of the flaw candidate. . When there are a plurality of applicable flaw candidates, information on the position is further added to make the information indicating the connection status of the flaws, and the plurality of flaws are associated.

The above description is for a case where a rectangular visual field is used, and in the invention of claim 1, the boundary area of the visual field uses only the width direction (regions d and h in FIG. 1). For the explanation of the image processing, it is easier to understand the rectangular visual field as the screen, so the rectangular visual field is hereinafter referred to as the screen.

Table 1 is a table showing an example of information indicating the feature amount of a flaw and the connection status of the flaw. These pieces of information correspond to FIG. The flaw (correctly a flaw candidate) G has no information indicating the connection status (or has connection information indicating that there is no possibility of connection) and is determined as an isolated flaw. The flaws A, C, and H have connection information in the boundary areas f, e, and h, respectively, and perform signal processing with an adjacent visual field to determine the flaw. here,
As the connection information, the boundary area symbols (a to h) and the coordinates in the screen are used.

[0022]

[Table 1]

FIG. 2 is a view showing each screen when the flaw extends over a plurality of screens. In this example, an overlapping area of each screen is set as a boundary area between adjacent screens. The flaw candidate A on screen 1 (same as FIG. 1)
Boundary region f has connection information, and flaw candidate B on screen 3 that follows this has connection information in boundary regions b and f (upper and lower sides in the figure).

Table 2 is a table in which information indicating the feature amount of the flaw and the connection status of the flaw for each screen shown in Table 1 are summarized. The above-mentioned flaw candidates A and B on the screens 1 and 3 are obtained from the positional relationship of the screen (front and rear in the longitudinal direction) and the connection information (the boundary area symbol and the position).
It is determined that the flaw is continuous. Since the flaw candidate B also has the connection information in the boundary area f, it is not determined as a flaw at this stage, and the flaw candidates are successively connected to flaw candidates on adjacent screens based on the connection information.

[0025]

[Table 2]

A flaw continuous in the width direction can be similarly detected. In this case, the connection status of the flaw is collected for the boundary region (regions d and h in FIG. 1) with the visual field of the adjacent light receiver. From the positional relationship of the screen (left and right in the width direction) and the connection information (symbol and position of the boundary area), it is determined whether or not the flaw is continuous.

If the screens overlap in the diagonal direction at the corners of the screens, if the screens overlap, the connection status of the four screens sharing the corners with the flaw candidates is collected. In Table 2, flaw candidates C on screens 1, 2, 3, and 4
For E, F, and D, connection information (boundary area symbol and position)
It can be seen that these flaw candidates are connected. Thereafter, by combining the feature amounts of the respective flaws, the flaws are detected as one flaw.

The method of detecting obliquely continuous flaws as described above can be expressed as a means for solving the problem. If information indicating the connection status at the boundary with the screen adjacent to the oblique direction is added, a plurality of methods can be used. The surface flaw detection method according to claim 2, wherein the surface flaw is detected based on the flaw information on the screen. Here, the diagonally adjacent screen is another screen sharing only one corner, and the boundary with the diagonally adjacent screen represents four corners of the screen.

In this manner, the flaw candidates are connected until there is no information indicating the connection status (or the content of the connection information becomes no possibility of connection), and thereafter, the flaw candidates are determined as flaws. Information can be obtained.

[0030]

FIG. 3 is an apparatus configuration diagram showing the configuration of an apparatus used as one embodiment of the present invention. In the figure, 1 is a light receiver, 2
Denotes a frame memory, 3 denotes a detection processing device, 4 denotes a comprehensive judgment device, and 5 denotes a final storage device. In this embodiment, the image processing unit is 512 pixels horizontally and 400 lines vertically.
Inspection of the running steel plate is performed by the two light receivers. A horizontal boundary region is set to 10 pixels, and a vertical boundary region is set to 10 lines. The border areas are overlapped to minimize oversight of flaws.

The signal from each light receiver 1 is imaged by the frame memory 2. When the frame memory 2 is full, the detection processing device 3 detects a flaw candidate from the screen, obtains the flaw characteristic amount and the connection information, and transmits them to the comprehensive judgment device 4. Two or more frame memories 2 are prepared for each light receiver, and are configured to store an image in another frame memory while detecting a flaw candidate in one frame memory. The overall determination device 4 determines a flaw based on the feature amount of the flaw candidate and the connection information. For flaw candidates without connection information,
The result is immediately displayed in one screen and transmitted to the final storage device 5.

FIG. 4 is a flowchart showing the contents of the connection process. For a flaw candidate having connection information, connection processing is performed as follows. In the connection process in the width direction, the presence / absence of connection information in the boundary regions (d and h regions in FIG. 1) of adjacent screens obtained at the same time is checked. As for the flaw candidates covering a plurality of screens, the type and degree of the flaw are determined from the characteristic amount of the flaw on each screen, and transmitted to the final storage device 5.

In the connection process in the longitudinal direction, the presence / absence of connection information in the boundary region (the b and f regions in FIG. 1) between the latest screen and the immediately preceding screen is checked. If the flaw candidates have been completed, the type and degree of the flaw are determined from the flaw characteristic amount on each screen, and transmitted to the final storage device 5, as in the case of the width direction. If the flaw candidate still has the connection information in the boundary area with the next screen (figure 1f), the determination of the flaw is carried over to the next cycle. However, the feature amount and the connection information of the flaw candidate up to the present time are transmitted to the final storage device 5.

The diagonal connection process is performed by checking for the presence or absence of connection information in the previous screen corner area (e and g areas in FIG. 1) and the latest screen corner area (a and c areas in FIG. 1). Examine
The whole image of the flaw candidate is determined. Thereafter, the type and degree of the flaw are determined from the characteristic amount of the flaw, and transmitted to the final storage device 5 as described above.

The final storage device 5 maps the flaw judgment result sent from the comprehensive judgment device 4 on a display and displays it, and records the result on a fixed recording medium. The recorded contents include not only the final determination result for the connected flaws, but also feature amounts and connection information obtained on all screens from the point of occurrence of the flaw to the point of termination. The adjustment of the flaw detection device is performed by collating the flaw detection result with the visual inspection. However, the device of the present invention stores the feature amount and the connection information for each screen, so that the target to be adjusted is narrowed down ( It is easy to perform a feature amount extraction process or a connection processing process, and it is possible to perform highly accurate detection.

[0036]

According to the present invention, various types of flaws extending over a plurality of screens can be correctly detected by detecting flaws using information indicating the connection state between adjacent visual fields together with the feature amounts of the flaws. be able to.

Further, even for a long flaw, since the characteristic amount and the connection information for each screen are stored, it is possible to adjust the detection device with high accuracy.

[Brief description of the drawings]

FIG. 1 is a diagram showing the form of a flaw at a visual field and its boundary.

FIG. 2 is a diagram showing each screen when a flaw extends over a plurality of screens.

FIG. 3 is an apparatus configuration diagram showing a configuration of an apparatus used as one embodiment of the present invention.

FIG. 4 is a flowchart showing the contents of a linking process.

FIG. 5 is an external view schematically showing a conventional technique.

[Explanation of symbols]

 ah boundary i inside Ah flaw

 ──────────────────────────────────────────────────の Continued on the front page (72) Inventor Takahiko Oshige 1-2-1 Marunouchi, Chiyoda-ku, Tokyo Inside Nihon Kokan Co., Ltd.

Claims (2)

[Claims] 1. A method of irradiating light to the surface of a band-like body and capturing the reflected light with a plurality of light receivers arranged in a width direction to detect surface flaws. With respect to the candidate, together with the feature amount of the flaw in the field of view, information indicating the connection status at the boundary with the field of view of the adjacent light receiver is added, and the surface flaw is detected based on the information of the flaws in a plurality of fields of view. Surface flaw detection method. 2. A method for irradiating light to the surface of a band, capturing the reflected light with a plurality of light receivers arranged in a width direction, and dividing the screen in a longitudinal direction to perform a surface flaw inspection. For the flaw candidates detected in the above, together with the feature amount of the flaw in the screen, information indicating the connection status at the boundary with the screen adjacent in the width direction and the longitudinal direction is added, and the flaws of the plurality of screens are added. A surface flaw detection method characterized by detecting a surface flaw based on information.