JPS5847015B2 - edge detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an edge detection device in a fluorescent magnetic particle flaw detection device, etc., and an object of the present invention is to enable accurate detection of edge portions of a test material.

Fluorescent magnetic powder liquid is sprinkled on the surface of the material to be inspected for flaws, the fluorescent magnetic powder is attached to the surface flaw due to the magnetization of the material to be inspected, and the fluorescent magnetic powder pattern is imaged by an imaging device to detect surface flaws. Fluorescent magnetic particle flaw detection devices have been known for a long time, but in this type of automatic surface flaw detection device, in order to determine the position of surface flaws existing on the surface of the test material, It is necessary to detect the edges of

Therefore, conventionally, visible light is irradiated onto the edge of the specimen from the same side as the imaging device, and the reflected light from the edge is imaged.

However, in this case, the brightness on the surface to be tested gradually decreases as it moves toward the edges, so the edges do not clearly appear as a clear change in brightness, and therefore accurate edge detection cannot be expected.

This is due to the fact that in the case of a test material such as a square billet, the edges thereof are curved surfaces.

The present invention solves these conventional problems, and its characteristics are as follows: An edge detection light source irradiates a light beam onto the edge of a material to be inspected, and the reflected light from the edge is detected. In an edge detection device that detects edges of a material to be inspected by capturing an image with an imaging device, an extension line connecting the imaging device and the observation limit point of the edge portion and a normal line of the edge portion passing through the observation limit point. The point is that a light source for edge detection is placed in between.

Hereinafter, the present invention will be described in detail with reference to the illustrated embodiments.
The figure shows an example of an automatic surface flaw detection device using the fluorescent magnetic particle detection method. In the figure, 1 is a test material such as a square billet,
The test material 1 is transported in the direction of the arrow a at a weight of approximately 4 to 30 g by a ■-shaped transport roller 3 and a pinch roller (not shown) with the adjacent flaw detection surface 2 facing upward. .

Reference numeral 4 denotes a preliminary water washer which is disposed laterally above the starting end of the specimen 1 and has a plurality of injection nozzles 5 at predetermined intervals in the lateral direction. This is to remove impurities such as dust, fine scale, oil, etc. adhering to the surface 2 to be tested by injecting pressurized water.

Reference numeral 6 denotes a first draining device for removing water adhering to each surface of the material 1 to be inspected after preliminary water washing, and includes four air injection nozzles corresponding to each surface of the material 1 to be inspected. Each of the air injection nozzles 7 is provided so as to inject compressed air from the outer side of the object 1 in the conveyance direction diagonally rearward with respect to each facing surface.

Reference numeral 8 denotes a magnetic particle sprayer for dispersing fluorescent magnetic powder liquid onto the surface 2 to be inspected of the test material 1, which is arranged horizontally above the test material 1 as shown in FIG. It has a plurality of spraying nozzles 9, and is configured so that the fluorescent magnetic powder liquid is evenly sprayed from each spraying nozzle 9 onto each surface 2 to be inspected of the material 1 to be inspected.

10 is a magnetization device that magnetizes the material 1 to be tested;
4 iron cores 11 arranged on the left, right, front and rear, four excitation coils 12 wound around each iron core 11, and each iron core 11.
Each excitation coil 12 is connected to a three-phase AC power source so as to generate a rotating magnetic field in the specimen 1.

The magnetization device 10 is mounted on a tracking device (not shown) that follows the material 1 to be inspected, and maintains the gap δ constant regardless of field conditions such as bending or positional fluctuation of the material 1 to be inspected. This ensures uniform magnetization over the surface.

Reference numeral 14 denotes a second draining device for removing moisture from the fluorescent magnetic powder liquid sprinkled on the surface 2 to be tested of the material 1 to be inspected, and as shown in FIG. A nozzle 15 is provided, and each air injection nozzle 15 is arranged so as to inject compressed air diagonally backward with respect to each flaw detection surface 2 from the outer side in the direction of conveyance of the material 1 to be inspected.

Note that this second draining device 14 can also use steam instead of compressed air.

Further, since this is provided for the purpose of removing moisture from the fluorescent magnetic powder liquid and drying it, it is also possible to use a heating method.

Reference numeral 16 denotes an imaging device such as an industrial television camera for taking an image of the surface 2 to be inspected of the material 1 to be inspected, and as shown in FIG. Each imaging device 16 is provided with an ultraviolet cut filter 18 to prevent ultraviolet rays directly reflected from the surface 2 to be detected from entering the lens 17.

Reference numeral 19 denotes a visible light source for edge detection that irradiates visible light for edge detection onto the edge portion 1a of the specimen 1, and is equipped with an incandescent light bulb 20 and a condensing lens 21.
In the illustrated example, reference numeral 9 indicates the specimen 1 on the opposite side of the imaging device 16.
They are arranged at lower positions corresponding to the edge part 22 of the edge part 22, but their arrangement positions are along the extension line 24 connecting the imaging device 16 and the observation limit point 23 of the edge part 22, as shown in FIG. , range 2 with the normal 25 passing through the observation limit point 23
It should be within 6.

Reference numeral 27 denotes a visible light source for large flaw detection that irradiates visible light for large flaw detection onto the surface 2 to be tested of the material 1 to be inspected. Two pieces are provided on both sides of the surface 2 with an inclination angle of about 12 degrees.

Reference numeral 30 denotes an ultraviolet light source that irradiates ultraviolet light onto the surface 2 to be inspected of the material 1 to be inspected, and includes a high-pressure mercury lamp 31 and a visible light cut filter 32.
and a condensing lens 33, each of which is provided corresponding to each surface 2 to be tested.

The field of view 34 of the imaging device 16 is as shown in FIG. 6, and within this field of view 34, as shown in the figure, there are visible light irradiation area 35 for edge detection, visible light irradiation area 36 for large flaw detection, and ultraviolet ray irradiation area 3.
7 are respectively set, and out of all the scanning lines 38 of the imaging device 16, scanning lines 39, corresponding to each irradiation area 35, 36, 37 are set.
40 and 41 are used for edge detection, large flaw detection, and small flaw detection, respectively.

Reference numeral 42 denotes a signal processing device that processes the video signal from the imaging device 16, and is configured as shown in the block diagram shown in FIG. Scratch removing device 45, marking device 46
etc. are connected.

In FIG. 7, 47 is a video amplifier that amplifies the video signal from the imaging device 16, 48 is an automatic sensitivity adjustment circuit, and 4
9 is a sample-and-hold A/D converter that samples a video signal and converts it into a digital signal; 50 is an accumulator;
51 is a D/A converter.

52 is a synchronization separation circuit that separates a vertical synchronization signal and a horizontal synchronization signal from a video signal, and 53 is a clock pulse generation circuit driven by the horizontal synchronization signal.

54 is an edge detection circuit that detects an edge signal corresponding to the edge portion 22 of the specimen 1 from the video signal, and this edge detection circuit 54 is an edge signal correction circuit that reuses the previous edge signal when the edge signal cannot be detected. It is equipped with

55 is a scanning line position specifying circuit for specifying a scanning line position; 56
is an edge detection scanning line setter for setting the edge detection scanning line 39; 57 is a large flaw detection scanning line setting device for setting the large flaw detection scanning line 40; and 58 is for setting the small flaw detection scanning line 41. This is a scanning line setting device for detecting small defects.

59 is an AND circuit, 60 is a sampling stress generation circuit, 61 is a material width setting device for setting the material width W1 of the material to be inspected 1,
Reference numeral 62 is a material width addition setting device that adds the width W2 from the clock pulse generation position to the edge signal detection position to the material width W1. This position becomes the actual starting position of the sampling palace.

63 is a vertical line extraction circuit that spatially differentiates the digital signal in the vertical direction to extract horizontal lines; a vertical line extraction circuit 64 that delays the digital signal by one scanning line; It comprises a vertical difference circuit 65 for differentiating signals, a vertical peak detection circuit 66, and a vertical buffer memory 67.

68 is a horizontal line extraction circuit that spatially differentiates the digital signal in the horizontal direction and extracts a line in the vertical direction; a horizontal line extraction circuit 69 that delays the digital signal by one bit; a horizontal-difference circuit 70 for differentiating the signal, and a horizontal-peak detection circuit 7
1 and a horizontal buffer memory 72.

Vertical-peak detection circuit 66 and horizontal-peak detection circuit 7
1, 4 bits of the differential signal constitute one word, and its peak value is taken as the representative value.

73 is a vertical local image memory that sequentially stores vertical differential signals corresponding to a required number of scanning lines (for example, 7) for each scanning line, and 74 stores horizontal differential signals in a required number of scanning lines. A horizontal-local image memory correspondingly stores sequentially for each scanning line, these vertical and horizontal-local image memos IJ
73 and 74 are 108-bit shift registers 75 to 8.
1 and 82 to 88 are connected in series.

Note that each shift register 75 to 81. and 82-88 are
Since the 4-bit peak value is used as the representative value in the previous stage, it actually corresponds to each of the four scanning lines.

89 is a line segment determination circuit that determines a line segment by adding the required number of stored signals in a predetermined line segment direction;
-97, the adders 90-97 add the signals stored in the vertical and horizontal local image memo 1J73.74 in eight directions, respectively, as shown in FIGS. and 82-88.

68 is a judgment circuit, and a threshold value setter 9 for setting a threshold value.
9, comparators 100 to 107 that compare the threshold value with the added value of each adder 90 to 97, and a peak detection circuit 108 that uses the peak value of each comparison value as a flaw signal.

109 is an oil detection circuit that discriminates the required number of bits of signals in the required number of scanning lines and generates an oil detection signal when all of them are 1; an integrating circuit 110 that integrates 4 bits of the digital signal; An oil detection threshold setting device 111 that sets a threshold, and a comparator 112 that compares the integral value and the threshold.
It consists of a 7-bit shift register 113, an AND circuit 114 that takes the conditions for its parallel output, 7-bit shift registers 115 to 121 connected in series, and an AND circuit 122 that takes the conditions for each output. ing.

Reference numeral 123 denotes a gate circuit that cuts off a flaw signal from the determination circuit 98 when an oil detection signal from the oil detection circuit 109 is received.

124 is a buffer memory, and 125 is a delay circuit.

In the above configuration, when making deep scratches on the surface of the test material 1 such as a square billet, first, the test material 1 is moved in the direction of the arrow a by the conveying roller 3 with the adjacent test surface 2 facing upward. Transport at a constant speed.

Next, pressurized water is injected from the spray nozzle 5 of the preliminary water washer 4 onto the surface of the material to be inspected 1 to remove dust, fine scales, Wash off impurities such as oil.

This prevents foreign matter from getting mixed into the fluorescent magnetic powder liquid and reducing the effective magnetic particle concentration when the fluorescent magnetic powder liquid is sprayed if foreign substances adhere to the surface of the material to be tested 1. It's for a reason.

The effective magnetic particle concentration refers to the concentration of fluorescent magnetic particles that exist in the liquid and are useful for flaw detection.The amount of fluorescent magnetic particles that adhere when the test material 1 is magnetized varies greatly depending on this effective magnetic particle concentration, Powder significantly reduces flaw detection performance.

In addition, although the fluorescent magnetic powder liquid is generally collected and reused, this prevents the detection ability from being degraded due to contamination with foreign matter.

After the preliminary washing is completed, compressed air is injected from each air injection nozzle 7 of the first draining device 6 onto the surface of the material 1 to be inspected.
Remove water adhering to the material surface during preliminary washing.

In other words, when water is mixed in the fluorescent magnetic powder liquid, its concentration decreases and the flow of the fluorescent magnetic powder liquid from the surface to be inspected 2 is accelerated. Remove as much water as possible to prevent the fluorescent magnetic powder liquid from flowing down.

Next, each spray nozzle 9 of the magnetic powder sprayer 8 sprays fluorescent magnetic powder at a rate of about 50 A' per minute onto the surface 2 to be inspected of the material 1 to be inspected.

At this time, the material 1 to be tested is being transported with its surface to be tested 2 facing upward, and each surface to be tested 2 is inclined at approximately 45 degrees. Compared to the type in which the material to be inspected is conveyed with the surface to be tested perpendicular, the flow of the fluorescent magnetic powder liquid on the surface to be tested 2 is slower, and the fluorescent magnetic powder liquid on the surface to be tested 2 is slower. The adhesion rate can be increased.

When the scattering of the fluorescent magnetic powder liquid is completed, the magnetizing device 10 applies a rotating magnetic field to the specimen 1 to magnetize it, and the fluorescent magnetic powder 126 scattered on the surface 2 to be inspected is applied to small flaws as shown in FIG. 9B. It is attached to the small flaw part 127 by leakage magnetic flux from 127, and after magnetization by this magnetization device 10, the second draining device 1
Compressed air is injected onto each surface 2 to be inspected by the air injection nozzles 15 of 4, and the fluorescent magnetic particles 1 adhering to the small flaws 127 are removed.
Remove as much water as possible, leaving only 26.

This is to increase the luminance of light emitted from the part where the fluorescent magnetic powder 126 is attached.

In other words, fluorescent magnetic powder liquid is made by suspending iron powder such as iron oxide with fluorescent paint in water, but it cannot be suspended simply by mixing it, so it is difficult to suspend it by mixing it with a dispersant. There must be.

However, commercially available dispersants strongly absorb ultraviolet light.

As an example, the absorption characteristics of the dispersant Cimarron 45-8 (trade name) manufactured by Shimadzu Corporation are shown.

FIG. 10 shows the spectral transmission characteristics of this dispersant Cimarron 45-8 at a concentration of 1% and a liquid thickness of 10 .mu.m, which shows that absorption occurs at wavelengths of about 390 .mu.m or less.

Therefore, usually, ultraviolet light from a mercury lamp is used for fluorescence excitation, but since its strongest luminous intensity is at 366.3 mμ,
Due to the dispersant, ultraviolet light may not be sufficiently transmitted.

FIG. 11 shows the change characteristics of luminance with respect to liquid thickness when fluorescent magnetic powder is actually excited.

However, the luminance is 1 when the dispersant is 0%.

In this way, the luminance was 80% when the dispersant was 1% (liquid thickness 1 μm), but when the dispersant was 2% (liquid thickness 1 μm), the luminance was 60%.
It decreases with age of about %.

Generally, about 1 to 3% of the dispersant is mixed into the fluorescent magnetic powder liquid.

Furthermore, when small flaws 126 are attached to the surface 2 to be tested as shown in FIG. 9B, a liquid (aqueous solution) 128 fills the spaces between the fluorescent magnetic particles 126, forming a relatively thick water film. Therefore, even if ultraviolet light is irradiated during flaw detection, the fluorescent magnetic particles 126 exposed on the surface will emit light, but the inside will not emit light, so the luminance will be about % to % of the dry state. Normal.

Thereupon, a fluorescent magnetic powder liquid is sprayed to magnetize the specimen 1 and the fluorescent magnetic powder 126 is attached to the small flaw 127 as shown in FIG. 9B. Air is injected onto the surface 2 to be inspected to remove moisture between the fluorescent magnetic particles 126 and dry them, thereby increasing the luminance of light emission.

In addition, since there is no leakage magnetic flux in large flaws 129 such as dents,
Fluorescent magnetic particles do not adhere magnetically, but depending on the depth, fluorescent magnetic powder liquid may remain as shown in Figure 9A, and this may have an adverse effect on detecting large defects using the reflective method. There is.

However, by injecting compressed air from the air injection nozzle 15 as described above, the fluorescent magnetic powder liquid adhering to the large flaw part 129 can be removed at the same time, improving the performance when detecting large flaws in the subsequent stage. do.

The compressed air may be either cold or hot, and the same effect can be obtained if water vapor is used instead.

Note that the magnetic powder liquid sprayer 8 and the magnetization device 10 may be arranged in opposite directions, and the fluorescent magnetic powder liquid may be dispersed after the specimen 1 is magnetized.

Next, each irradiation area 35° 36 .
37, each light source 19, 27 as shown in FIG. 1 and FIG.
, 30, and the imaging device 16
In this case, the position of the visible light source 19 for edge detection is set as follows.

In Fig. 12, the normal 0- to the cylindrical surface S of radius r
Assume that observation is made at point P on point P, and a parallel ray of intensity Io is irradiated from a direction forming an angle α with this point.

Consider the brightness of point Q on cylindrical surface S observed from point P (however, when viewing cylindrical surface S from point P, the left side of point R cannot be seen).

The amount of light incident on the surface element rdφ at point Q is It is.

When light is reflected from this surface, the angular dependence of the reflected light is f(
θ).

When observing this reflected light at point P, the amount of reflected light per unit area is: If the cylindrical surface S is a perfect isotropic scatterer (approximately correct if the surface is quite rough), Next, the brightness on the material surface when the angle φo-75 degrees is set and the direction α of the incident light incident on the edge portion 22 is changed every 30 degrees within the range of 130 to 120 degrees as shown in Fig. 5. Figure 13 shows the change in height.

a is α=−30 degrees, and in this case, it gradually becomes darker as it moves toward the edges, and the edges are unclear and cannot be determined.

Further, it is not possible to completely irradiate the observation limit point 23 when looking at the edge portion 22 side from the imaging device 16.

b indicates α20 degrees, that is, illumination from the front; in this case, the light gradually becomes darker toward the edges, and the edges are unclear.

C is α230 degrees and d is α=60 degrees, and in these cases, the highest intensity point is created before reaching the observation limit point 23.

e indicates the case where α=90 degrees, and f indicates the case where α is 2120 degrees; in these cases, the light gradually becomes brighter toward the observation limit point 23, and suddenly becomes darker after the observation limit point 23.

The light beam from the visible light source 19 for edge detection is reflected and changed by the edge part 22 in this way, so in order to clearly recognize the edge, it must gradually become brighter toward the observation limit point 23 as shown in e and f above. It is most desirable that the light suddenly become dark after point 23.

In other words, for edge detection, it is most advantageous to arrange the edge detection visible light source 19 within the range 26 and irradiate the edge portion 22 side with a light beam from within this range 26, and moreover, reliable edge detection is possible. .

As described above, the edge irradiates visible light from the edge detection visible light source 19 into the edge detection visible light irradiation area 35, and uses the reflected light reflected by the edge portion 22 of the specimen 1 as the edge detection scanning line 39. The large flaw part 129 irradiates the visible light irradiation area 36 with visible light from the visible light source 27 for large flaw detection to detect the large flaw part 12 as shown in FIG. 9A.
9 is captured and detected by the large flaw detection scanning line 40, and the small flaw portion 127 irradiates the small flaw detection ultraviolet ray irradiation area 37 with ultraviolet light from the ultraviolet light source 30.
The fluorescent magnetic powder 126 attached to the small flaw 127 is caused to emit light, and the fluorescent light beam is captured and detected by the small flaw detection scanning line 41.

On the signal processing device 42 side, first, the number of scanning lines is set according to each detection purpose using an edge detection scanning line setting device 56, a large flaw detection scanning line setting device 57, and a small flaw detection scanning line setting device 58. This is designated by the scanning line position designation circuit 55.

A video signal obtained by imaging the inspection surface 2 of the test material 1 with the imaging device 16 is amplified by the video amplifier 47, and then separated into a vertical synchronization signal and a horizontal synchronization signal by the synchronization separation circuit 52. A clock pulse generation circuit 53 operates according to the synchronization signal and generates a clock pulse.

On the other hand, if the video signal at this time corresponds to the edge detection scanning line 39 set by the edge detection scanning line setter 56, the edge detection circuit 54 operates to detect the edge of the specimen 1. do.

Then, since the width W2 from the clock pulse generation position to the edge signal detection position is known, the workpiece width setter 61 sets the workpiece width W1, and the workpiece width addition setter 62 sets the width W2.

Therefore, the actual sampling pulse generation position is located back from the edge portion 22 by the width W2, and the sampling pulse generation circuit 60 generates sampling pulses from this position at predetermined time intervals.

The video signal passed through the video amplifier 7 is applied to an automatic sensitivity adjustment circuit 48, where a predetermined sensitivity adjustment is performed.

This automatic sensitivity adjustment circuit 48 integrates the total amount of emitted light at a specific position within the ultraviolet irradiation area 37 using an accumulator 50, converts this into an analog signal using a D/A converter 51,
Variable gain control is performed using the analog signal to change the amplification factor of the video signal.

The video signal whose sensitivity has been adjusted through the automatic sensitivity adjustment circuit 48 enters a sample-and-hold A/D converter 49, where it is sampled by a sampling pulse from a sampling pulse generation circuit 60 and converted into a digital signal.

That is, the video signal becomes a 432-bit digital signal obtained by dividing the material width W1 of the material to be inspected 1 into 432 parts by this sample-hold A/D converter 49, and is sent to the vertical and horizontal line extraction circuits 63 and 68 at the next stage. Sent.

In the horizontal line extraction circuit 68, the digital signal is sequentially inputted to the horizontal difference circuit 70, which passes through the horizontal delay circuit 69 and makes a difference with the delayed signal delayed by 1 bit. - The peak value of the 4 bits of the differential signal is determined by the peak detection circuit 71, and this peak value is used as the representative value of the 4 bits.

Therefore, from now on, there will be a 108-bit signal for one scanning line, and each bit of the signal, that is, the representative value, will be sequentially stored in the horizontal buffer memory 72.

In the vertical line extraction circuit 63, a vertical difference circuit 65 subtracts the digital signal corresponding to the scanning line from the delayed signal delayed by one scanning line via the vertical delay circuit 64, and outputs the 4-bit difference signal. A representative value of .times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..times..

In other words, these vertical and horizontal line extraction circuits 63 and 68 spatially differentiate the digital signals and extract the light emitting point on the surface 2 to be inspected of the material 1 to be inspected [this is the reflected light from the large flaw 129 or the small flaw This corresponds to the emitted light beam of the fluorescent magnetic powder 126 attached to the fluorescent magnetic powder 127] and its surroundings, and the location where the change is the most drastic is selected.

Therefore, for example, when a test pattern 130 whose entire elliptical surface emits light as shown in FIG. 14 is imaged, by combining the signals passed through the vertical and horizontal line extraction circuits 63 and 68,
The image becomes an elliptical image 131, and all internal light-emitting parts are removed.

The vertical and horizontal local image memories 73 and 74 connected to the next stage of the vertical and horizontal line extraction circuits 63 and 68 each have one
08-bit seven shift registers 75 to 81 and 82
-88 are connected in series, and the signals that have passed through the vertical and horizontal line extraction circuits 63 and 68 are sent to these shift registers 75.
~81. and 82 to 88 are sequentially stored corresponding to each scanning line.

Each signal is sent to each adder 90 to 97 of the line segment determining circuit 89.
are outputted in parallel, and added in eight directions by the separate adders 90 to 97 to determine the line segment.

In other words, as shown in FIGS. 8A to 8H, each adder 90 to 97
detects connections between the signal sides by adding 7 mesh signals vertically and horizontally in respective predetermined directions.
In the case shown in FIG. 4, each adder 90-97 determines the continuity of the signal as shown on the right side thereof.

A predetermined threshold value is set in the threshold value setting device 99 in advance,
Comparators 100 to 107 compare the added value obtained by adding 7-bit signals in a predetermined direction with each adder 90 to 97 and the threshold value, and if the added value is less than the threshold value, there is no continuity in the predetermined direction. Only when the peak value is above the threshold value, a signal is sent to the peak detection circuit 108, and a flaw signal indicating that there is a surface flaw having continuity in a predetermined direction with the peak value as a representative value is generated.

When transporting the test material 1, it is necessary to apply lubricating oil or the like, and although this lubricating oil or the like is washed off during preliminary water washing, it may still remain on the surface of the material.

If oil remains on the surface 2 to be inspected, the surface may reflect ultraviolet rays and emit fluorescent light, and may be detected as if there is a surface flaw.

Therefore, an oil detection circuit 109 is provided so that when the entire fixed area of the surface 2 to be detected is in a light emitting state, this is detected as oil.

That is, the 432-bit digital signal from the sample-and-hold A/D converter 49 is integrated by the integrating circuit 110 to obtain 10
It is converted into an 8-bit signal, and each signal is compared with the threshold value from the oil detection threshold value setting device 111 in a comparator 112.

Then, the signals from the comparator 112 are sequentially sent to the shift register 113, and an AND condition of the parallel output cuff bits is obtained by an AND circuit 114.

Next, the data is sent to each of the 7-bit shift registers 115 to 121, and an AND condition for the outputs of each of the shift registers 115 to 121 is determined by an AND circuit 122.

Therefore, in this case, if all 7-bit signals in the vertical and horizontal directions are 1, an oil detection signal is generated from the AND circuit 122, and it is assumed that the light emitting part on the surface 2 to be tested is oil, and the oil detection signal causes the gate circuit 123 to This cuts off the flaw signal from the judgment circuit 98.

If there is a flaw signal, the marking device 46 is driven via the delay circuit 125 to directly mark the surface flaw portion of the material 1 to be inspected, or the marking device 46 is driven via the buffer memory 124 to directly mark the surface flaw portion of the specimen 1.
It is also possible to automatically remove the flaw by driving 5.

It is also possible to display the information on the monitor television 43 or record it on a recording paper using the recorder 44, and either of these may be adopted as necessary.

It goes without saying that the imaging device 16 and the light sources 19, 27, and 30 are attached to the tracking device.

As described in detail in the embodiments above, in the present invention, the edge detection light source is arranged between the extension line connecting the imaging device and the observation limit point of the edge portion and the normal line of the edge portion passing through the observation limit point. ,
Since this light source is configured to irradiate the edge of the material to be inspected with light, the brightness of the surface to be micro-inspected changes to a doggy pattern as it moves toward the edge, and reaches its maximum at the observation limit point. .

Therefore, since the observation limit point clearly appears, when it is imaged with an imaging device, the edge can be detected accurately, and the effect is extremely significant.

[Brief explanation of the drawing]

The drawings show one embodiment of the present invention; FIG. 1 is a side view of the whole, FIG. 2 is a view taken along the line X-X in FIG. 1, and FIG. 3 is a view taken along the Y-Y line in FIG. 1. 4 is a view along the Z-Z line in FIG. 1, FIG. 5 is an explanatory diagram showing the direction of light irradiation on the edge portion, FIG. 6 is an explanatory diagram of the field of view and scanning line of the imaging device, and FIG. The figure is a block diagram of the signal processing device, Figure 8 is an explanatory diagram for line segment determination, and Figure 9 is a block diagram of the signal processing device.
Diagram A. B is a cross-sectional view of the surface flaw, FIG. 10 is a spectral transmission characteristic diagram of the dispersant, FIG. 11 is a luminance characteristic diagram of fluorescent magnetic powder, FIG. 12 is an illustration of light reflection, and FIG. 13 is an edge part. FIG. 14 is an explanatory diagram of the signal processing process. 1... Material to be inspected, 2... Surface to be inspected, 4.
...Preliminary water washer, 6...First drainer,
8...Magnetic powder liquid spreader, 10...Magnetizing device, 14...Second draining device, 16...
Imaging device, 19... Visible light for edge detection, source;
22...Edge section, 42...Signal processing device, 64...Vertical line extraction circuit, 68...
. . . horizontal line extraction circuit, 73 . . . vertical local image memory, 74 . . . horizontal local image memory,
89... Line segment determination circuit, 98... Judgment circuit, 109... Oil detection circuit, 123...
...Gate circuit.

Claims (1)

[Claims] 1. In an edge detection device that detects edges of a material to be inspected by emitting a light beam from an edge detection light source to the edge portion of the material to be inspected and capturing an image of the reflected light beam from the edge portion with an imaging device, the edge detection device includes an imaging device and An edge detection device characterized in that an edge detection light source is disposed between an extension line connecting an observation limit point of an edge portion and a normal line of the edge portion passing through the observation limit point.