JPH031619B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a method for detecting surface defects in a material to be tested, and particularly to a method for non-contact detection of surface defects in a red-hot steel slab in a continuous casting process.

Conventional structure and its problems There are two methods for detecting surface flaws on red-hot objects: contact type and non-contact type.
Two methods have been proposed. However, in the contact type, since the target object is at a high temperature, heat resistance of the contact part is required, and flatness of the target object becomes a problem, and sufficient detection accuracy cannot be expected. Therefore, a non-contact detection method is generally used to detect surface flaws on a red-hot object.

Optical methods are generally used to detect surface flaws in a non-contact manner, and the following two methods have been proposed so far.

One method is to irradiate a red-hot object with light in the visible or ultraviolet range, capture a shadow image of the surface of the red-hot object, and detect defects from this image signal.

The other method is detection based on the self-luminescence of a red-hot object. This is a method that uses the temperature difference between the surface of a red-hot object and the inside of the red-hot object exposed by a flaw to take an image in the infrared region, and detects a flaw from this video signal.

However, with the above two methods, the former may lead to erroneous judgments, such as assuming that if there are irregularities other than flaws on the surface of the red-hot object, it is also a flaw, and the latter may lead to erroneous judgments, such as considering the unevenness on the surface of the red-hot object as a flaw. This method has the disadvantage that a sufficient signal-to-noise ratio cannot be obtained by using this method alone, since non-uniformity of surface temperature occurs due to this.

Furthermore, Japanese Patent Application Laid-Open No. 131192/1983 proposes a method in which the above two methods are implemented in one optical system and each flaw is detected using a separate system depending on the type of flaw. However, this method still includes the drawbacks of the two optical detection methods described above and is not a substantial improvement. For example, when the surface of a red-hot object has irregularities that are not flaws, irradiation with light in the visible or ultraviolet range produces shadows from the irregularities, and this method identifies this as a flaw. Furthermore, the signal-to-noise ratio of the video signal has not been improved by self-emission.

In addition, JP-A-52-139482 and JP-A-55
-160837 also obtains binary flaw signals from each of the above two detection methods and calculates the logical product of these signals, so it still contains the drawbacks of the above two methods and is not an essential improvement. It hasn't become familiar.

Purpose of the Invention The present invention aims to solve the problems of the optical detection method described above. A detection device is provided.

Structure of the Invention The apparatus according to the present invention captures an image of the surface of a red-hot object in one image.
The projected image is projected by two imaging optical systems.
The optical paths are separated, one of the passing wavelength ranges is a near-infrared range, and the other passing wavelength range is a shorter wavelength range than the above-mentioned near-infrared range, and one of the photoelectric signals of the image pickup devices located on both projection planes is This is a surface flaw detection device for a red-hot object, which has a signal processing section that divides one by the other. Furthermore, this device has an illumination device that emits spectral radiation in the near-infrared region, which is one of the above-mentioned passing wavelength regions.

With the above configuration, the long wavelength side of the near-infrared region is used to detect the shadow image of a flaw on a red-hot object using illumination, and the detection of self-luminescence of the red-hot object is performed on the short wavelength side of the visible range with a good signal-to-noise ratio. .

Furthermore, by performing binarization processing on the quotient obtained by dividing one of the photoelectric outputs of both obtained in this way by the other, it is possible to generate more flexible defects than when each is binarized independently. Discrimination can be realized.

DESCRIPTION OF THE EMBODIMENTS FIG. 1 shows an example of a surface flaw detection device of the present invention, which is used as a surface flaw detection device for hot steel slabs in steel plants, particularly continuous casting equipment. In this case, the surface temperature of the slab is approximately 800-900°C, and the internal temperature is 1300°C.
As shown in Fig. 1, this device includes an illumination device 2 that emits mainly spectral radiation in the near-infrared region that illuminates a red-hot object, that is, a hot slab 1, and an image pickup device that images the surface of the slab illuminated by this illumination device. device 3 and a signal processing unit 10 that processes signals from the imaging device
It consists of

The illumination device 2 is configured by, for example, a combination of a light source such as an incandescent lamp that emits spectral radiation also in the infrared region, and a sharp cut filter.

In the imaging device 3, 4 is a projection optical system for projecting the surface of the hot slab 1. 5 is a half mirror placed on the image field side, which divides the optical path into two.
The image plane is divided into two parts to form two imaging planes. 7 and 9
are image sensors arranged on the two imaging planes.
A sharp cut filter 8 is arranged in the optical path of the image sensor 9 so as to have sensitivity in the near-infrared region, and a sharp cut filter 8 is arranged in the optical path of the image sensor 7 so as to have sensitivity mainly in the short wavelength side of the visible region. A cut filter 6 is arranged.

FIG. 2 shows a block diagram of the signal processing unit 10, in which 7 and 9 are image pickup elements, 33 and 34
35 is a gain adjustment circuit for adjusting the gain and linear characteristics of the photoelectric signal of the image sensor; 35 is a division circuit for dividing one of the photoelectric outputs from the gain adjustment circuits 33 and 34 by the other and calculating the quotient; 37 compares the output of the division circuit with the reference voltage generator 36,
A binarization circuit 38 represents a flaw discrimination signal produced by the binarization circuit 37.

The operation of the flaw detection apparatus of this embodiment configured as described above will be described below.

The image sensor 9 captures a shadow image of the hot slab 1 caused by the illumination light from the illumination device 2 . That is, the concave portion on the surface of the hot slab 1 becomes a shadow, and the photoelectric signal corresponding to one line of scanning of the image sensor takes the form shown at 17 in FIG. 3b. Further, the image sensor 7 images the self-luminous pattern of the hot slab 1.
That is, if there is a flaw 11 on the surface of the hot slab 1 in FIG. 3, the high temperature part inside the slab 14 is exposed, so that a flaw signal 22 as shown in c can be obtained.

In general, all objects can be said to radiate thermal energy determined by their own temperature and emissivity. This amount of radiant energy, a black body (emissivity is 1)
In the case of , it is expressed by the following formula.

Meλ=C ₁ λ ₁ λ ⁵ / (expλT/C ₂ −1) [WCm ^-2 μ ^-1 ] …(1) Where, Meλ is the spectral radiant emittance, T is the absolute temperature of the perfect radiator, and λ is The wavelength, C ₁ and C ₂ can be expressed as constants.

In Figure 3, the temperature inside the hot slab 14 is
Assuming that the temperature of the slab surface 13 is 1300℃ and 850℃,
When formula (1) is illustrated, a spectral radiation distribution as shown in FIG. 4 is obtained.

In FIG. 1, when a solid-state image sensor made of silicon is used as the image sensor, the sensitivity wavelength range of this element is 0.4 μm to 1.2 μm. Therefore, the sensitivity wavelength range of the image sensor 7, which also takes into account the characteristics of the infrared cut filter 6, is 0.4 μm to 0.7 μm. In other words, as can be seen from Figure 4, by using the short wavelength radiation of the self-luminous radiation of slab defects, for example in the visible range of 0.7 μm or less, a sufficient SN ratio can be achieved compared to that in the near-infrared region. It is possible to obtain a photoelectric signal with

In addition, a sharp cut filter 8 (for example,
R-73 filter that passes light of 0.73μm or more)
The sensitivity wavelength range of the solid-state image sensor 9 is 0.73 μm to 1.2 μm, taking into consideration the characteristics of . On the other hand, by using the above-mentioned R-73 filter as the sharp cut filter of the illumination device, the image pickup device 9 can directly capture a shadow image depending on the wavelength band of the illumination light. In this case, since self-luminous radiation of 0.73 μm to 1.2 μm will also be detected at the same time, the output of the optical radiation from the illumination light should be sufficiently large and the level ratio with the self-luminous radiation should be about 10:1 to 20:1. By taking this, the influence of self-luminous radiation can be removed.

Further, by using an incandescent light bulb such as a halogen light bulb in the lighting device, stable illumination light can be obtained regardless of the scanning frequency of the detector, and efficient illumination light can be obtained in the near-infrared region. In addition, there is no fluctuation in illumination light that is noticeable due to movement of the bright spot inside the arc tube as in a discharge lamp.

As shown in the above embodiment, the self-luminous component of a slab flaw is detected at a range of 0.4 μm to 0.7 μm, and the shadow image of a slab flaw by an illumination device is detected at a range of 0.7 μm to 0.7 μm.
By using the 1.2 μm wavelength range, it is possible to obtain flaw signals with a good S/N ratio in both wavelength ranges.

The wavelength setting of the sharp cut filter used in the illumination device 2 and imaging device 3 is 0.73μ.
By using a sharp cut filter on the wavelength side longer than 0.73 μm, both photoelectric signals can be shifted to the short wavelength side and the long wavelength side with m as the center. can be balanced.

Further, by using a red reflective dichroic mirror instead of the half mirror 5, crosstalk between light incident on both light receivers can be reduced, and a flaw signal with a good signal-to-noise ratio can be obtained.

Next, the operation of the signal processing section will be explained based on FIG. Hot slab 1 with a cross section as shown in Figure 3a
As mentioned above, the photoelectric signal when the image is taken is the third one.
The signals are shown in Figures b and c. In addition to the flaw signal 17 caused by the flaw 11 on the hot slab 1, the photoelectric signal 15 of the image sensor 9 that captures a shadow image includes a signal 16 caused by a recess on the hot slab surface, and a signal 16 caused by the recess on the hot slab surface. There are signals 18, 19 produced by the upper high and low reflectors. In addition, the photoelectric signal 21 of the image sensor 7 that images the self-luminous pattern includes:
In addition to the flaw signal 22 due to the flaw 11 on the hot slab 1, there is a signal 23 caused by the irregular temperature distribution on the slab surface.

If the above two signals are individually binarized, 16, 17, and 19 of the photoelectric signal 15 and 22 and 23 of the photoelectric signal 21 are recognized as defect signals, and the logic of these two signals is Even if logical product is taken, let alone sum, the signal is easily determined to be a defect because it is generated by signals other than defects, such as 19 and 23.

In this signal processing, as shown in FIG. 2, the level of the photoelectric signals obtained by the image sensors 7 and 9 is adjusted in gain adjustment circuits 33 and 34 based on the average signal voltage of the portion without flaws. The signals of the gain adjustment circuits 33 and 34 are divided by one signal by the other signal in a division circuit 35.

The signal output 25 shown at d in FIG.
This shows the result of dividing 1 by the photoelectric signal 15.
Although the crack signal 27 is emphasized from the two photoelectric signals, the other photoelectric signal waveforms 16, 18, 19, and 23 are all canceled out.

The principle of the above determination method will be explained based on FIG. FIG. 5a illustrates the flaw discrimination method shown in the conventional example. When the shadow light decreases below Q or the self-emission exceeds P, it is determined that there is a flaw. In other words, it is shown that the area other than the area I is determined to be a flaw.

On the other hand, in Fig. 5b, the area above the curve connected by P'Q', that is, the area I, is not determined to be a flaw, and the area below this P'Q', that is, the photoelectric output due to self-luminescence, is shaded. This indicates that a defect is determined when the output of the division circuit for dividing the light by that of the light exceeds a certain value.

When comparing a and b in FIG. 5, in the present invention, areas 39 and 40 in area I are determined to be free of defects,
This indicates that the area No. 41 is determined to be a flaw. Applying the measurement example in FIG. 4 to FIG. 5, it can be seen that 16 and 19 of the photoelectric signals of shadow light belong to region 39, and 23 of the photoelectric signals of self-emission belong to region 40. The flaw determination method is
This shows that it is possible to make more flexible decisions than the conventional example. That is, since the gain of the photoelectric signal of self-luminous light and shadow light can be changed by the gain adjustment circuits 33 and 34, the judgment curve P'Q' having arbitrary linear characteristics within a certain limit is determined according to the actual measurement data. can be set, and a flexible algorithm can be created when determining defects.

Note that in the above embodiment, one threshold divides the output of the division circuit for both photoelectric signals into two.
Although the image is converted into values, a multi-value conversion circuit system may be arranged in place of the binarization circuit 37, and the area portions may be divided according to the type of flaw. In addition, in order to prevent the determination level shown in FIG. 5b from changing due to a decrease in the photoelectric output due to dirt on the projection optical system, etc., the output signals of the gain adjustment circuits 33 and 34 are memorized as appropriate, and this information is also used. By providing a function to automatically shift the reference voltage of the reference voltage generating section 36, stable determination operation can be maintained over a long period of time.

Effects of the Invention The device for detecting surface flaws on a red-hot object according to the present invention detects self-luminescence of a red-hot object in the short wavelength range of visible radiation,
By capturing shadow light caused by flaws in the near-infrared region, these two
By inputting two photoelectric signals to a division circuit through each gain adjustment circuit and binarizing the output of this division circuit, it is possible to set a flaw judgment standard according to the actually measured flaw judgment data, which makes it possible to accurately and flexibly Therefore, it is possible to provide a method for determining flaws with a high degree of practicality, and its practical value can be said to be extremely large.

[Brief explanation of the drawing]

FIG. 1 shows a principle diagram of an illumination device, an imaging device, and a signal processing section of the configuration according to the present invention. Figure 2 is a block diagram of the signal processing section, Figure 3a is the shape of the hot slab surface,
3b to d are the photoelectric signals obtained corresponding to the above a, FIG. 3e is the flaw detection signal, and FIG. 4 is the spectral radiation distribution of the slab surface calculated based on equation (1). FIG. 5A shows a conventional example of the flaw signal detection principle, and FIG. 5B shows a flaw detection principle according to the present invention. DESCRIPTION OF SYMBOLS 1...Hot slab, 2...Illuminating device, 3...Imaging device, 4...Projection optical system, 5...Half mirror, 7...Imaging element, 8...Sharp cut filter, 9...Imaging element , 10... signal processing section.

Claims (1)

[Claims] 1. An incandescent light bulb, an illumination device having means for selecting near-infrared light from the output light of the incandescent light bulb, and imaging optics for projecting the light selected by the illumination device onto a red-hot object. an optical element for obtaining two systems of imaging planes from this imaging optical system, a filter for passing radiation in the near-infrared region, which is provided separately on each of the optical paths divided into two systems, and the near-infrared region. a filter that passes visible radiation with a shorter wavelength than the outer region; an imaging device located on each of the imaging planes of the two systems; a means for dividing one of the photoelectric signals from the imaging device by the other; and the dividing means. A device for detecting surface flaws on a red-hot object, comprising a signal processing unit for binarizing the output of a red-hot object.