JPS6226417B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention provides mechanical parts such as bearing rollers,
The present invention relates to an automatic visual inspection device for cylindrical objects that continuously inspects the appearance of the outer peripheral surface of cylindrical objects such as uranium berets for nuclear fuel.

Conventionally, automatic visual inspection equipment for the outer circumferential surface of cylindrical objects
As described in Japanese Patent Application Laid-Open No. 52-85877, a cylindrical object mounted on two rotationally driven rollers is attached to a chain conveyor device at a predetermined interval, and a cylindrical object is mounted on two rotationally driven rollers, and A member extending upward is brought into contact with the bottom surface of the cylindrical objects that are fed in sequentially, and each cylindrical object is sequentially rotated around its axis and moved at a constant speed in the axial direction to perform stationary optical exploration. The device irradiates the outer circumferential surface of the cylindrical object with a spot of light, scans the outer circumferential surface of the cylindrical object in a spiral shape with a width of 1 mm, and captures the reflected light to detect chipping marks and other dents such as flaws. It was designed to test. However, with this conventional automatic visual inspection device for the outer circumferential surface of a cylindrical object, the outer circumferential surface of a cylindrical object with a length of 10 mm to 15 mm is inspected by scanning a minute width of 1 mm or less in a spiral shape. 10 to 15 cylindrical objects
The first drawback is that the inspection of one cylindrical object cannot be completed unless it is rotated more than once, resulting in slow inspection speed and inefficiency. Furthermore, as mentioned above, since each cylindrical object is rotated around its axis and moved in the axial direction, it is difficult to keep the rotational speed of the cylindrical object constant in the strict sense, making it impossible to inspect uniformly and with high precision. It has a second drawback. Therefore, in order to solve the second drawback, a large number of cylindrical objects arranged in a line were mounted on two rotationally driven rollers as described in Japanese Patent Application Laid-Open No. 125057/1983. The cylindrical object is rotated at a constant speed without being moved toward the heart, and the imaging device is composed of a line scanner, television camera, etc. that has at least a one-dimensional scanning area slightly larger than the length of the cylindrical object. The movement of the cylindrical object is stopped intermittently at intervals corresponding to the length of the cylindrical object (stepwise movement stop) in the axial direction of the cylindrical object, and during this stop, images are taken of all the cylindrical surfaces of the cylindrical object one after another to detect chips, cracks, etc. Devices for testing are known. However, this device also has the drawback of inefficiency because the imaging device is moved linearly intermittently at intervals corresponding to the length of the cylindrical object.

An object of the present invention is to eliminate the drawbacks of the prior art described above, and to move an imaging device that performs one-dimensional scanning and imaging and a large number of cylindrical objects arranged in a line in the axial direction of the cylindrical objects in a relatively continuous manner. The outer circumferential surface of a cylindrical object can be efficiently inspected while the cylindrical object is in the same state, and the appearance of each cylindrical object can be inspected with high accuracy by correcting any cumulative errors caused by variations in scanning speed in two-dimensional directions or the resolution of the encoder. An object of the present invention is to provide an automatic appearance inspection device for a cylindrical object.

That is, in order to achieve the above object, the present invention includes at least two rollers that rotate a large number of cylindrical objects at a constant speed about their axes in a state in which they are brought into contact with each other and aligned in a line. a rotation means, an imaging optical system that forms an image of the outer peripheral surface of the cylindrical object rotated by the rotation means, and an image formed by the imaging optical system one-dimensionally in the axial direction of the cylindrical object. An imaging device configured with a linear image sensor having a field of view width longer than the length of each cylindrical object, and a movement in which the imaging device and the plurality of cylindrical objects are moved relative to each other in the direction of the rotation axis in order to scan and image the plurality of cylindrical objects. means, an encoder for detecting the amount of movement by the moving means, and a first scanning line ys on the outer peripheral surface of the cylindrical object detected by the encoder in advance.
The edge position of the cylindrical object is determined from the reference position based on the amount of movement q from the reference position to the second scanning line yt separated by a large number of scanning lines M, and the video signal obtained by the linear image sensor by imaging the first scanning line ys. distance to
The distance _{e 1 from} the reference position to the edge position of the cylindrical object (e ₂
-e ₁ ), and a cumulative error measuring means that calculates the number of scanning lines Ns converted into an integer by dividing the number M of scanning lines by the error between coordinate setting means for setting coordinates in the scanning line direction accordingly and correcting the set coordinates for each number of scanning lines Ns to correct the error calculated by the cumulative error measuring means; edge detection means for detecting the starting end position and end position of the cylindrical object based on the video signals of each scanning line obtained and whose coordinates are set by the coordinate setting means; A binarization circuit converts the video signal of each scanning line into a binary pixel at a predetermined threshold value, and a binarization circuit converts the video signal of each scanning line obtained from the binarization circuit into a binary pixel for the entire circumference of the cylindrical object. a cutting means for cutting out an area of each cylindrical object based on the starting end position and ending position detected by the edge detecting means; The present invention is an automatic visual inspection apparatus for cylindrical objects, characterized in that it is equipped with a processing determining means for storing a binary pixelated signal in a storage means and determining a processing for inspecting the external appearance of each cylindrical object.

That is, the moving speed of the imaging device usually varies slightly, causing the image to be distorted as shown in FIG. 2, making it impossible to accurately determine the appearance of the circumferential surface of the cylindrical object. Therefore, it is necessary to accurately know the movement of the field of view due to movement of the imaging device. If the field of view moves,
The position of the external image of the cylindrical object within the field of view moves, and the amount of movement is determined by detecting the amount of movement of the imaging device.
It can be determined by multiplying by the imaging magnification. Therefore, when the imaging device moves a distance corresponding to one pitch of the light receiving surface of the imaging device, the image on the cylindrical surface shifts by one element, that is, one pixel number, on the light receiving surface. However, in reality, depending on the accuracy of the imaging device's movement amount detector, the imaging magnification error (difference between the calculated magnification and the actual magnification), etc., when converting the amount of movement of the imaging device to the above movement amount. , a slight error will occur. In particular, it is difficult to determine the exact value of the imaging magnification of an imaging device due to focus adjustment and other factors. This error usually accumulates during the imaging of one cylindrical object and becomes a non-negligible amount.

Therefore, the number of scanning lines required for the cumulative value of the shift error to reach an integer number of picture elements is measured in advance, and the video signal obtained from the imaging device is converted into binary picture elements by a binarization circuit. By the signal
Each time the imaging device scans the measured number of scanning lines, the integer number of picture elements and target coordinates are corrected and stored in a memory, and the two are read out from this memory.
The outer peripheral surface of the cylindrical object aligned in a line is sequentially inspected according to the pixelated signal.

The present invention will be specifically described below based on embodiments shown in the drawings. FIG. 1 is a perspective view showing a schematic configuration of an embodiment of an automatic visual inspection apparatus for cylindrical objects according to the present invention. 1 is a holding block supported by chain conveyors 2 installed on both sides. The holding block 1 rotatably supports rollers 4a and 4b that carry a large number of cylindrical objects 3 in a row and rotate them at a constant speed by contacting them. The motor 6 is connected to the motor 6 via a The holding block 1 has a loading V-groove 7, which allows a large number of cylindrical objects 3 to be moved in the axial direction and loaded onto the rollers 4a and 4b at the loading position, as shown in FIG. After the inspection of the outer circumferential surface of the cylindrical objects 3 is completed at the inspection position shown in FIG. A V-groove 8 for unloading is formed so that it can be moved and discharged. The holding block 1 is moved intermittently by the chain conveyor 2, and is accurately positioned and stopped at an inspection position by a positioning means (not shown). 9 is supported so as to be slidable in the horizontal direction along guide shafts 12 and 13 supported by side frames 10 and 11 installed at the inspection position, and is connected to a drive motor 14 via a speed reducer (not shown). Connected feed screw 15
A moving table configured to mesh with the
Illumination means 18 consisting of an illumination light source 16 and an irradiation lens 17 are provided on both sides of the lower end of the movable table 9.
a and 18b are installed, and the outer circumferential surface of the cylindrical object 3 is configured to be tilted from at least two opposing directions and irradiated with light. At the center of the movable table 9, an imaging lens 19 and an imaging device 5 for scanning and imaging at least one-dimensionally are provided. The imaging device 5 can be configured by, for example, a TV camera, a linear image sensor (one-dimensional solid-state imaging device), a two-dimensional image sensor, or the like. However, it is better to use a linear image sensor than to use a TV camera because of the high resolution.

However, at a loading position (not shown), a large number of cylindrical objects 3 are laterally fed and mounted in a line through the loading V-groove 7 of the holding block 1. Next, the chain conveyor 2 is moved intermittently by a predetermined amount and stopped, and the holding block 1
is positioned at the inspection position directly below the moving table 9 and stopped. Next, drive the motor 6 to
a and 4b are rotated at a constant speed, and a large number of cylindrical objects 3 closely aligned in a line are simultaneously rotated at a constant speed. When a large number of cylindrical objects 3 begin to rotate at a constant speed after a predetermined period of time has elapsed, the motor 14 is driven and the feed screw 15 moves the moving table 9 from the right end as shown in FIG. Continuously move to the left. The reason why the cylindrical object 3 is rotated only by the rollers 4a and 4b without being moved in the axial direction is to minimize fluctuations in the rotational speed of the cylindrical object 3 so that the imaging device 5 can detect the cylindrical object. This is to enable precise inspection by scanning and imaging the outer peripheral surface of No. 3 with uniform density. Moreover, in order to significantly improve efficiency by continuously inspecting a large number of cylindrical objects 3, the cylindrical objects are arranged on a movable table 9 equipped with an imaging device 5 and illumination means 18a and 18b, as described above. It was made to move continuously in the axial direction. However, even if the moving table 9 is moved continuously, there are some fluctuations in the moving speed, so the imaging device (linear image sensor) 5 has a curved field of view S as shown in FIG. The pattern is moved diagonally and imaged. The imaging device 5 repeatedly performs one-dimensional scanning imaging based on the scanning synchronization signal Hsync. Note that a one-dimensionally scanned and imaged video signal 21 is continuously read from the imaging device 5 (FIG. 3).

Next, a device for inspecting the appearance of a cylindrical object based on this video signal 21 will be described with reference to FIG. That is, 20 is a rotary encoder connected to the left end of the feed screw 15 as shown in FIG. 1, and outputs a pulse signal every time the moving table 9 is moved by a predetermined amount. That is, the rotary encoder 20 detects the feed speed of the moving table 9. 22 is a synchronization signal generator which generates a clock signal and a scanning synchronization signal Hsync. 23 is an A/D conversion circuit that converts the video signal 21 into a digital signal. Reference numeral 24 denotes an edge detection threshold setting circuit, which compares and searches the maximum value Vd of one scanning line of the digital video signal output from the A/D conversion circuit 23 using a digital comparator, and sets the maximum value Vd in advance. This circuit sets the first edge detection threshold value Vth1 by subtracting the value ΔV. Reference numeral 25 denotes a digital binary circuit for converting the digital video signal output from the A/D conversion circuit 23 into a binary picture element according to the edge detection threshold value Vth1 set by the first edge detection threshold setting circuit 24. It is a conversion circuit. Reference numeral 26 denotes a coordinate control circuit that controls coordinates so that the screen imaged by the imaging device 5 remains stationary based on the pulse signal output from the rotary encoder 20, the clock signal output from the synchronization signal generator, and the scanning synchronization signal Hsync. (coordinate setting means). This coordinate control circuit 26
creates an edge detection signal that detects that the first edge position BEG1 has been reached when the first edge position BEG1 coordinate signal output from the first edge detection circuit 27 is detected and the scanning synchronization signal Hsync is input. After receiving the inspection completion signal 29 from the circuit 28 indicating that the inspection of the entire outer peripheral surface of one cylindrical object has been completed, when the edge detection signal generation circuit 28 outputs the first edge position signal, the coordinate update load signal is output. A coordinate update specifying circuit 30 to output, and an adder 3 to add the value (x _r+ 〓) to the coordinate (address) x _H of the left end of the visual field before updating as shown in FIG.
1, a subtracter 32 that subtracts the initial edge position coordinate BEG1 from the output of the adder 31 [x _H + (x _r+ 〓)], and when a coordinate update load signal is output from the coordinate update designation circuit 30, A scan start coordinate counter 33 that loads the output of the subtracter 32 and subtracts it with a pulse output from the rotary encoder 20 to count the scan start coordinate (address) at the left end of the visual field, and a coordinate signal of the scan start coordinate counter 33. The coordinate counter 34 receives the coordinates, adds the number of clock pulses to the coordinates, and extracts the coordinates on the scanning line. 27 calculates the difference (differentiation) of the binary pixelized signal obtained from the binarization circuit 25 as shown in FIG. As shown in Figure 4c, the frequency distribution is obtained for, for example, 256 scanning line segments, and this frequency distribution is successively compared using a comparator to determine the starting edge position BEG1 showing the maximum and the ending edge position BEG1. This is the first edge detection circuit that outputs the edge position BEG2 in coordinate values (address). These edge detection threshold setting circuits 2
4, the digital binarization circuit 25, and the first edge detection circuit 27 constitute edge detection means. 3
5 is the first digital comparator that outputs a signal to start extracting the low level Vp that determines the second threshold when the output coordinates of the coordinate counter 34 reach the updated coordinates EG1 of the first edge position; Reference numeral 36 denotes a final digital comparator 40 which outputs a signal to end extraction of the low level Vp and start extraction of the high level Vd2 when the output coordinates of the coordinate counter 34 reach the updated coordinates EG2 of the final edge position. is a digital comparator that outputs a signal to end the extraction of high level Vd2 when the coordinate Ed1' of the first edge position of the updated previous cylindrical object is reached. 37 is a signal obtained from the first digital comparator 35 and the last digital comparator 36, and as shown in FIG. The output digital video signal is stored in the address of the first harpoon corresponding to the brightness (output level), the frequency distribution of the brightness of the outer peripheral surface area of the cylindrical object is determined, and the maximum frequency value Vp is determined. Furthermore, as shown in FIG. 4d, the digital video signals output from the A/D conversion circuit 23 for 255 scanning lines in the range EG2 to EG1' are stored in a second memory corresponding to the brightness (output level). Store it in the address, find the frequency distribution of brightness in the edge area, find its maximum frequency value Vd2, Vth2 = α (Vd2 - Vp) +
This is a second threshold setting circuit that sets a threshold value Vth2 called Vp. 38 is A/A by the second threshold Vth2 set by the second threshold setting circuit.
This is a digital binarization circuit that binarizes the digital video signal output from the D conversion circuit 23. 3
9 is a noise cancellation circuit. 41 is a comparator that calculates EG1+δ from the EG1 signal and forms a signal that is “1” from this, and 42 is a comparator that calculates EG2−δ from the EG2 signal and forms a signal that is “1” so far. 43 is one of the binary pixel signals output from the noise canceling circuit 39 (EG
This is a gate circuit that allows the region from 1+δ) to (EG2−δ) to pass through. These comparators 4
1 and 42 and the gate circuit 43 constitute a cutting means. Reference numeral 44 denotes a processing judgment circuit that stores the binary pixel signal that has passed through the gate circuit in a memory having, for example, 512 addresses, and determines the area of the defect, the outline of the defect, etc., and judges pass/fail. 45 is the second
A test start signal forming circuit 46 outputs a test start signal when a signal with a second threshold determined from the threshold setting circuit 37 is input; This is an inspection end signal forming circuit that counts the number of lines and generates an inspection end signal when it reaches a predetermined value (the number of lines that scan the entire outer circumference of the cylindrical object). 50 is a cumulative error measuring circuit, which includes a setting circuit 51 for setting the number of scanning lines, and a scanning synchronization signal Hsync.
A second scanning line is separated from the first scanning line ys by the number M of scanning lines set by the setting circuit 50.
A counter 52 counts the pulse signals output from the rotary encoder 20 up to yt, and a digital 2 counter 52 counts the pulse signals output from the rotary encoder 20 until
The distance _e1 from the scanning synchronization signal to the first rising edge of the binarized signal obtained from the digitizing circuit 25 is determined by pixel counting through a clock pulse, and then after the number of scanning lines M set by the setting circuit 51, The distance e ₂ from the horizontal synchronizing signal to the first rising edge of the binarized signal obtained from the digital binarization circuit 25 on the second scanning line yt of ₂ - e ₁ ), and a comparator that calculates the difference between the value (e ₂ - e ₁ ) obtained by this arithmetic circuit 53 and the value q (corresponding to the target coordinate) obtained by the counter 52. 5
4 and the number of scanning lines M set in the setting circuit 51 by the value output from the comparator 54 and rounding off to the nearest integer to obtain the number N _s of scanning lines.
and a division circuit 55 for extracting the .

56 stores the positive and negative signals outputted from the division circuit 55 of the cumulative error measuring circuit 50 and the number of scanning lines _Ns , and a positive signal is output every time the scanning synchronization signal Hsync reaches the number _Ns of scanning lines. If a negative signal is input, the last pulse (integer picture element number) of the pulses input from the rotary encoder 20 to the coordinate correction circuit 57 is erased, and if a negative signal is input, the coordinate correction circuit 57 is input from the rotary encoder 20. This is a control circuit that operates the coordinate correction circuit 57 so as to add one pulse (an integer number of picture elements) to the last of the pulses input to the coordinate correction circuit 57.

With the above configuration, the cylindrical objects 3 arranged in a line are rotated at a constant speed by the rollers 4a and 4b as shown in FIG.
6 is turned on and the motor 14 is driven to move the moving table 9 from the end. Then, while moving in the direction of the arrow shown in FIG. 1 together with the moving table 9, the imaging device 5 one-dimensionally scans and images the surface of the cylindrical object 3 over the field of view S as shown in FIG. A video signal is obtained and converted by the A/D conversion circuit 23 into a digital video signal shown in multiple gradations. Therefore, since it is initially in the reset state, the scanning start coordinate counter 33 has a value x _r +ε set in advance.
is loaded, and as the imaging device 5 moves, a pulse signal is input from the rotary encoder 20 and sequentially subtracted, and the scanning start coordinate is stored. Therefore, even if the moving speed of the imaging device 5 fluctuates, the pulse signal from the rotary encoder 20 (one pulse is output every time the imaging device 5 moves one pixel) is used as shown in FIG. is corrected linearly.
That is, as shown in Fig. 5, the X coordinate of the left edge of the visual field in the scanning line of ye is x _o , but as a result of detecting that the visual field has moved by one pixel before the start of scanning of ym, the starting point of scanning of ym is The address of becomes x _o -1. However, the coordinate counter 34 receives the scanning synchronization signal.
The scan start point coordinate x _o stored in the scan start point coordinate counter 33 is loaded every Hsync, and the coordinates of each scanning line are determined by counting clock signals from the coordinate x _o .

That is, x _r is set so that the X-axis origin is located on the left outer side of the field of view of the imaging device. For example, the number of picture elements in the field of view of the imaging device may be S, and the update may be performed first before x _r +ε does not jump out from the right edge of the field of view. The updating method is to set xr (origin position) + ε to the left edge of the visual field when the cylindrical object does not exist within the visual field and xr+ε is shifted by a certain amount from the left edge of the visual field. These operations are repeated to search for the position of the image of the cylindrical object, and when the first edge BEG1 of the cylindrical object is detected, x _r (origin position) is moved to a position closer to it by a distance of ε.

If the video signal is read by approaching the image of the cylindrical object and setting x _r in this way, the width of the image data memory will be as shown in Fig.
As shown in K), it is only necessary to add some margin to the width of the image of the largest cylindrical object on the right side from x _r .

In other words, the memory only needs to have a width of x _r to x _r +ε+L+K. Each time an image of a cylindrical object appears, this narrow memory moves and reads the image, making it possible to use the memory efficiently and shorten the time required for pattern processing.

On the other hand, the first edge detection threshold setting circuit 2
4 searches for the maximum value by comparing it with a digital comparator etc. in a predetermined width region where the digital signal first changes from high level to low level, and subtracts the preset value △V from this maximum value Vd. Then, the first threshold value Vth1 is determined. In response to the signal of the first threshold value Vth1, the digital binarization circuit 25 converts the digital video signal into binary picture elements and inputs it to the first edge detection circuit 27.

By the way, as shown by the solid line in FIG.
The amount of movement L _s (=e ₂ −
e ₁ ) and the target coordinate q, which is set while being shifted based on the pulse train signal detected from the rotary encoder 20, as shown by the dotted line in FIG. Therefore, before inspection, the number M of scanning lines is set in the setting circuit 51, and the number M of scanning lines is set in the setting circuit 51.
is the digital 2 in some first scanning line y _s
The binary pixel signal outputted from the digitization circuit 25 is received, and the number of clock pulses gated by the distance e ₁ to the rising edge is calculated using a counter or the like based on the scanning synchronization signal Hsync. The binary pixel signal output from the digital binarization circuit 25 is received at the second scanning line y _k separated by the set number of scanning lines M, and the distance e ₂ to the rising edge is calculated based on the scanning synchronization signal Hsync. The number of gated clock pulses is counted with a counter, etc., and e ₁ is subtracted from e ₂ to find the movement amount L _s (= e ₂
-e ₁ ), and the counter 52 calculates the rotary encoder 20 from the first scanning line _ys to the second scanning line _yt .
Find q by counting the pulse signals output from
A comparator 54 compares the pattern movement amount L _s obtained by the arithmetic circuit 53 with the target coordinate shift amount q obtained by the counter 52 to obtain the accumulated error in M scanning lines _. /q
is calculated to determine whether it is positive or negative, and the number of scanning lines N _s converted into an integer. In other words, dividing by the reference coordinate shift amount q is the number of scanning lines N _s where an error of one picture element occurs.
He sought. However, the coordinate correction circuit 57 adds or removes one pulse from the pulse signal output from the rotary encoder 20 so as to erase the error of one pixel in a cycle of _Ns scanning lines, and adjusts the scanning start coordinate counter. 33, the scanning start coordinate counter 33 corrects the scanning start coordinate (address) of the left end of the visual field to be shifted by one address (one pixel) at a cycle of _Ns , thereby preventing an increase in error. . In this way, before starting the inspection, the first scanning line y
Information on the surface pattern of the cylindrical object from _s to the second scanning line y _t is taken in, and the number N _s of scanning lines determined by the cumulative error measuring circuit 50 is stored in the control circuit 56 . Then, the imaging device 5 is moved from the right side to the left side in FIG.
When images are sequentially taken and inspected, the shift error is corrected at a period of _Ns , and the solid line and dotted line shown in FIG. 6 can be made to almost match. Although it has been described that the pattern and the target coordinates always match when there is no shift error, in reality it is difficult to match the shift timings of both, and therefore a mismatch of about ±1 picture element occurs.

Next, the first edge detection circuit 27 calculates the difference (differentiation) of this binary pixelized signal, stores the falling edge in the falling counter, and stores the rising edge in the rising counter. From the time when the pixelized signal of the scanning line starts to include the rising edge BEG1, calculate the frequency distribution for H = 256 scanning lines, for example, and calculate the frequency distribution for falling edges stored in the falling counter. Read the distribution from the first address to a predetermined address (x _r + ε + L/2), compare it successively with a comparator, output the address showing the maximum (first edge position coordinate BEG1), and store it in the rise counter. Read out the frequency distribution for rising edges from the address (x _r + ε + L/2) to the address (x _r + ε + 3/2 L), perform successive comparisons with the comparator, and output the address indicating the maximum (last edge position coordinate BEG2). do. coordinate counter 3
When the coordinate value from 4 becomes the coordinate value of this BEG1, the comparator 35 is activated, the edge detection signal generation circuit 28 is activated, and the coordinate update designation circuit 30 outputs the coordinate update load signal along with the scan synchronization signal Hsync, and the adder 31 is the scanning start coordinate counter 33
The contents of the scan start coordinate counter 33 are added to the coordinate values x _H and (x _r +ε) output from the subtracter 32, and the value x _I = (x _H + x _r +ε−BEG1) is subtracted by BEG1 by the subtracter 32. is updated. This is x _I −x _r
This is from the relationship: = x _H - (BEG1 - ε). Then, the initial value of the coordinate counter 34 is also updated, and the first
The address of the edge detection circuit 27 is also updated. Then, the second threshold setting circuit 37 adjusts the brightness level of the digital video signal in the range of EG1 to EG2 by the updated and outputted signals of EG1, EG2, and EG1' from the comparators 35, 36, and 40. Store it in the corresponding memory address and find the frequency distribution,
Find the brightness level Vp that shows the maximum value, and similarly
Frequency is stored in another memory address corresponding to the brightness level of the digital video signal in the range of EG2 to EG1' (when EG1' is not extracted, the search is performed with a predetermined width η set from EG2). The distribution is determined, the brightness level Vd2 showing the maximum value is searched, and the calculation Vth2=α(Vp−Vd ₂ )+Vd ₂ is performed to set the second threshold value Vth ₂ . Then, the digital video signal is converted into binary picture elements at a second threshold value Vth ₂ , and is composed of a 3×3 picture element array, and noise is eliminated by a noise canceling circuit 39 that takes the AND of all the picture elements. do. This binary picture element signal is the signal EG1+ from the comparators 41 and 42.
The signal passes through the gate circuit 43 between δ and EG2-δ and is input to the processing determination circuit 44. Processing determination circuit 44
For example, the size of the defect over the entire surface is calculated by counting the number of picture elements corresponding to the defect from the time when the test start signal is received from the test start signal forming circuit 45 until the test end signal is received from the test end signal forming circuit 46. Determine the condition of the defect by calculating or differentiating it in two dimensions to determine the outline picture element of the defect, count this to determine the defect outline length over the entire surface, and compare it with the judgment criteria to determine pass/fail. A test result signal 47 is output. In this way, while constantly correcting fluctuations in the moving speed of the imaging device 5, the coordinates at which the video signal is stored in the memory are shifted to follow the movement of the imaging device 5, and the coordinates at which the video signal is stored in the memory are updated one by one for each cylindrical object. The imaging device 5 can image the entire outer circumferential surface of each cylindrical object one by one in the same way as when it is stationary with respect to a large number of aligned cylindrical objects 3, thereby accurately and efficiently capturing images of the cylindrical objects. The appearance can now be inspected.

In particular, in order to uniformly scan the entire outer circumferential surface of a large number of aligned cylindrical objects with scanning lines, the row of cylindrical objects is simply rotated by a roller at a constant speed, and the row of cylindrical objects is slid on the roller in the axial direction. Instead, the imaging device was moved continuously in the axial direction of the cylindrical object to inspect the entire outer circumferential surface of a large number of aligned cylindrical objects.

As explained above, according to the present invention, a large number of cylindrical objects are arranged side by side on rollers that are rotated at a constant speed and rotated at a constant speed, and the imaging device and the large number of cylindrical objects are relatively arranged. move continuously in the direction of
The amount of movement (velocity) is detected, and the pixel coordinates of the video signal obtained from the imaging device are sequentially corrected and shifted to follow this movement so as to avoid cumulative errors. Since the coordinates are updated every second, this method is more efficient than the conventional method in which the imaging device is sent intermittently and images are taken only in a stationary state.Moreover, for a large number of arranged cylindrical objects, the encoder can be updated for each cylindrical object. It has a remarkable effect of correcting the cumulative error based on the resolution and inspecting the outer peripheral surface with high precision.

[Brief explanation of the drawing]

FIG. 1 is a perspective view showing a mechanical part of an automatic appearance inspection device for a cylindrical object according to an embodiment of the present invention, and FIG. 2 is a state in which the outer peripheral surface of the cylindrical object is imaged by the imaging device shown in FIG. 1. FIG. 3 is a schematic configuration diagram showing an embodiment of a device for inspecting the appearance of a cylindrical object based on the video signal obtained from the imaging device shown in FIG. 1, and FIG. 5 is a diagram showing signal waveforms obtained from the device and circuit shown in FIG. The figure shows the error between the amount of movement of the pattern on the cylindrical surface and the target coordinates. Explanation of symbols, 3... Cylindrical object, 4a, 4b...
roller, 5...imaging device, 9...movement table,
14... Drive motor, 15... Feed screw, 20...
...Rotary encoder, 23...A/D converter,
24...first threshold setting circuit, 25, 38...
... Digital binarization circuit, 26 ... Coordinate control circuit, 33 ... Scanning start point coordinate counter, 30 ...
Update designation circuit, 34... Coordinate counter, 27...
First edge detection circuit, 37... Second threshold setting circuit, 43... Gate circuit, 44... Processing inspection circuit, 45... Inspection start signal generation circuit, 46...
...Inspection end signal creation circuit, 50...Cumulative error measurement circuit, 51...Setting circuit, 52...Counter, 5
3... Arithmetic circuit, 54... Comparator, 55...
. . . Division circuit, 56 . . . Control circuit, 57 . . . Coordinate correction circuit.

Claims (1)

[Claims] 1. A rotating means having at least two rollers for rotating a large number of cylindrical objects at a constant speed around their axes while bringing them into contact with each other and aligning them in a line; An imaging optical system that images the outer peripheral surface of a cylindrical object, and an image formed by the imaging optical system is scanned one-dimensionally in the axial direction of the cylindrical object. An imaging device configured with a linear image sensor having a longer field of view, a moving means for relatively moving the imaging device and the plurality of cylindrical objects in the direction of a rotation axis, and detecting the amount of movement by the moving means. an encoder, a movement amount q from a first scanning line ys on the outer circumferential surface of a cylindrical object to a second scanning line yt separated by a large number M of scanning lines, which is detected in advance by this encoder, and the first scanning line Based on the video signal obtained by the linear image sensor by imaging ys, the distance e ₂ from the reference position to the edge position of the cylindrical object and the video signal obtained by the linear image sensor by imaging the second scanning line yt. Cumulative error of calculating the number of scanning lines Ns converted into an integer by dividing the number of scanning lines M by the error between the distance e ₁ from the reference position to the edge position of the cylindrical object and the distance (e ₂ - _e 1 ) The measuring means and the coordinates in the scanning line direction are set according to the amount of movement between each scanning line sequentially detected by the encoder, and the error calculated by the cumulative error measuring means is to be corrected for each number of scanning lines Ns. a coordinate setting means for correcting the set coordinates, and a starting end position of the cylindrical object based on video signals of each scanning line sequentially obtained from the linear image sensor and whose coordinates are set by the coordinate setting means. and an edge detection means for detecting the terminal position and the end position;
The digitization circuit and the binary pixelization signal of each scanning line obtained from the binarization circuit are measured over the entire circumference of the cylindrical object, based on the starting and ending positions detected by the edge detection means. a cutting means for cutting out a region of each cylindrical object, and a storage means storing a binary pixelized signal of the region of each cylindrical object cut out by the cutting means, and inspecting the appearance of each cylindrical object. 1. An automatic appearance inspection device for a cylindrical object, characterized in that it is equipped with processing determining means for determining whether the object should be processed.