JPS622260B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION When detecting flaws on a surface to be inspected, especially flaws on the surface of a slab, by line scanning, the present invention eliminates fluctuating signals at edges using a single scanning detector, and detects flaws existing near edges. This invention relates to a signal processing method that can accurately detect scratches caused by scratches.

Conventionally, infrared detection devices have been used in the steel industry to detect flaws in heated slabs, steel plates, and the like. In this case, a detection signal due to a steep temperature change appears at the edge portion, but this is originally of the nature to be cut. Moreover, it must be distinguished from other flaws in its vicinity.

Two methods are known as conventional methods for detecting this type of flaw. The first method, as shown in Figure 1a, is to move the slab 1 in the longitudinal direction 2, place two scanning detectors 3 ₁ and 3 ₂ directly above it, and scan the right and left sides of the center line. share. That is, the scanning angle θ
₁ and θ ₂ , the respective scanning directions 4 ₁ and 4 ₂ are reversed, and the beginning portions of the scans are overlapped to a point beyond the right end R and the left end L. A common method for this scanning is to scan and condense infrared rays at a scanning point by reflecting them with a rotating polygon mirror having a rotation axis parallel to the center line of the slab. A flaw signal is extracted from the temperature signal waveform obtained by this scanning and focusing through a CFAR circuit shown in FIG. In this CFAR circuit, an input waveform is split into two through an amplifier 11, one is passed through a low-pass filter LPF 12 to make it a smooth waveform, and the other original signal is passed through a delay circuit 13 that gives the same delay as the amount of delay of LPF 12, and both of these waveforms are is input to the differential amplifier 14 to obtain a difference waveform. Among the fine change signals obtained as a result, a particularly prominent peak signal is detected as a flaw signal. Figure b shows the detection signal e _s1 between the center C and the right edge R detected by the scanning detector ₃₁ , and figure b shows the difference signal waveform e ₀₁ obtained by passing this through the CFAR circuit. Similarly, the figure b shows the detection signal e _s2 between the center C and the left edge L;
This is the signal waveform e ₀₂ obtained through the CFAR circuit. As can be seen in Figure b, since a large fluctuation waveform occurs at startup and at the edges, scanning is started overlapping at the center and cut at the center point. Cut. As a result, only the flaw signal can be extracted.

The second method, as shown in Figure 2a, is to place one scanning detector 3 directly above and scan an area larger than the slab width at a scanning angle θ using the same scanning method as described above to measure the temperature. The signal waveform is processed by the digital circuit method shown in FIG. The circuit shown in c in the figure is a digital circuit similar to the CFAR circuit described above, which converts the sampled input signal into a digital signal by the A/D converter 15, and divides the scanning section into three sections: the left and right ends and the middle section. The two ends are set as narrow average sections where the level changes quickly, and a narrow average calculation circuit 16 calculates the average level that changes quickly, and the middle section is set as a wide average section, and the wide average calculation circuit 17 calculates the average level that changes slowly.
Turn each on/off in synchronization with the scanning timing.
The difference between the signal and the original signal is determined by the difference calculation circuit 19 after the signal is turned off using the switch circuit 18 .

Although the operation of this method is shown in analog waveforms for simplicity in FIG. 2B, it is equivalent to a digital waveform. In other words, the signal waveform e _S detected by the running difference detector 3 in figure b is divided into three sections, the both ends and the middle part, as described above in figure b.
By switching the section average level, fluctuations at both ends are reduced, and the flaw signal in the middle becomes more prominent. As a result, only the flaw signal in the middle part can be extracted as shown in FIG.

In these two signal processing methods, the first
The second method requires the use of two scanning detectors, which has the disadvantage of complicating the entire detection system. Since it is included in the value, the difference component may become small and may not be detected.

SUMMARY OF THE INVENTION An object of the present invention is to provide a signal processing method that can remove fluctuating signals at edges and accurately detect flaws near edges using a single scanning detector.

In order to achieve the above object, the signal processing method of the present invention scans a surface to be inspected with a scanning line of a predetermined length and detects a signal of a flaw existing on the surface, and includes a memory for one scanning line, After storing the signal for one scanning line, the second half including the center of a predetermined length is read out in the forward direction, and the first half including the center is read out in the reverse direction, and each is read out as a smoothed signal separately. The present invention is characterized in that the flaw signal output obtained by cutting the signal before the center point address and the signal after the address of the cut point of the edge portion set in advance is extracted.

The present invention will be described in detail below with reference to examples.

In the present invention, as in FIG. 2a, the scanning detector 3 is
The scanning method is the same as described above, that is, the infrared rays at the scanning point are reflected by a rotating polygon mirror to scan the entire width of the slab, and the infrared light is collected by an infrared detector for detection. The detection circuit is the CFAR circuit of FIG. 1c or a similar digital circuit. What is particularly provided in the present invention is a memory for storing temperature signal waveforms for one scanning line, means for controlling writing and reading of this memory at predetermined timing, and an output for outputting as a series of scratch signals. It's memory.

FIGS. 3a to 3e are explanatory views of each stage in the embodiment of the present invention.

Figure a shows the detection stage, and figure a shows the scanning line corresponding to the time axis t, and the scanning line is intermittently generated by reflection from the rotating polygon mirror. That is, the scanning line from the left end L of the slab 1 to the right end R via the center line C by a certain mirror surface is the effective period of scanning,
When scanning is performed at a constant angular velocity, the slab surface is scanned in a tan wave shape with the center as the origin, as shown in figure a, and the signal outputs e, _S , E, etc. in figure a correspond to the slab position of each scanning line. is detected. At time _tB , the scanning line disappears from the right end of the slab, and there is a scanning blank period until the scanning line reappears at the left end due to the next mirror surface. In the present invention, during this time, the memory contents are read out and unnecessary signals such as edge portions are removed.

Figure b is the writing stage, and as shown in figure b, the slab position in figure a is set as an address in the memory for one scanning line, and as shown in figure b,
A to E, etc. of the signal output e _S are written.

Figure c shows the memory read stage. Reading from the memory is performed during the blank period of the scan described with reference to FIG. That is, as shown in FIG.
is read in the reverse direction from the center as shown in c in the same figure, and the second half 2 including the center of the memory is read out in the reverse direction from the center.
is read out from the center in the forward direction as shown in FIG.

d in the figure is a stage of detecting a flaw signal and processing the signal. That is, the memory read contents 1 and 2 read out in FIG. 1c are respectively input to the CFAR circuit in FIG.
The flaw signals 1 and 2 shown in FIG. d are output, and among these, the signal before the center point address and the signal after the address of the cut point of the edge section set in advance are cut off. And the remaining scratch signals 1 and 2
The flaw signal 1 is stored in the output memory for one scanning line in the reverse direction and the flaw signal 2 is stored in the forward direction starting from the center point using the output memory write address shown in FIG. As a result, unnecessary fluctuation signals at the initial stage and edge portions are removed and only necessary flaw signals are stored in the output memory.

Note that the signal at the center point of one of the scratch signals 1 and 2 is stored in the center point address.

Finally, e in the figure shows the flaw signal readout stage.
According to the output memory read address shown in the figure e, the time and the distance on the slab maintain a linear relationship and are read out as the actually necessary flaw signal shown in the figure e.

As described above, according to the present invention, the memory for one scanning line is provided, and after storing the signal for one scanning line, the second half including the center part of a predetermined length is forwardly moved to the center part. The first half including the edges is read out in the opposite direction, each edge is processed separately, and the flaw signal output is extracted. In other words, using one scanning detector, the same procedure as in the case of two scanning detectors is performed using memory, and fluctuating signals at both edges can be cut with high precision. , it is possible to correctly extract flaws in the vicinity. In this way, unnecessary signals are removed, and only the necessary flaw signals are stored in the output memory for one scanning line and taken out as a series of flaw signal outputs, which can be used for quality inspection of the slab surface.

[Brief explanation of the drawing]

FIGS. 1 a to 2 c and 2 a to c are explanatory diagrams of the configuration and operation of the conventional example, and FIGS. 3 a to e are explanatory diagrams of the procedure of the embodiment of the present invention. In the figures, 1 is a slab; 2 is a moving direction, 3 is a scanning detector, 11 is an amplifier, 12
is a low-pass filter, 13 is a delay circuit, 14 is a differential amplifier, 15 is an A/D converter, 16 is a narrow average calculation circuit, 17 is a wide average calculation circuit, 18 is a switch circuit, and 19 is a difference calculation circuit. show.

Claims (1)

[Scope of Claims] 1. A signal processing method that scans a surface to be inspected with a scanning line of a predetermined length and detects signals of flaws existing on the surface, which includes a memory for one scanning line and stores signals for one scanning line. After storing, the second half including the center portion of a predetermined length is read out in the forward direction, and the first half including the center portion is read out in the reverse direction. A signal processing method characterized by extracting a flaw signal output obtained by cutting a signal before a point address and a signal after an address of a cut point of a preset edge portion. 2. Both flaw signal outputs obtained by processing each edge separately are sequentially stored in an output memory for one scanning line in reverse and forward directions starting from the center point, and read from the beginning as a series of flaw signal outputs. The signal processing method according to claim 1, characterized in that the signal processing method performs extraction.