JPH04238256A - Defect section image detecting method for continuously running sheet-shaped body - Google Patents

Abstract

PURPOSE:To grasp the pattern of a defect section as a two-dimensional static image in the continuously running state at a high speed and analyze it in detail by providing at least two kinds of first-in first-out type image memories, and utilizing them in turn. CONSTITUTION:A continuously running sheet-shaped body 1 is illuminated by a light source 2, the pattern is grasped by a one-dimensional image sensor camera 3, a defect section is detected 4 from the video signal outputted repeatedly, and it is image-processed 5 to detect a defect section image. The video signal is converted into the digital image data by an A/D conversion section 5a with high-speed conversion performance. At least two first-in first-out type image memories 5a1, 5a2 recording the data and having the memory capacity to express the two-dimensional image of the defect section are provided. When the defect sections occur continuously, one of the image memories 5a1, 5a2 records the data, the other having recorded the data transfers the data to multiple storage image memories 5b1-5b8, both of them are utilized in turn to grasp multiple defect section pattern static images, and they can be analyzed in detail and coped with by digital image process calculation.

Description

[Detailed description of the invention]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting images of defects in a continuously running sheet-like object.

2. Description of the Related Art Regarding a method for detecting images of defective parts of a moving sheet-like object, as shown in Japanese Patent Application Laid-Open No. 60-207004, it is difficult to detect defects in a moving sheet-like object in the case of a low-speed moving production line (for example, several tens of meters per minute or less). There are examples in which the video signal output from a one-dimensional image sensor camera is displayed on a monitor via a signal processing circuit, an image memory, and a display control circuit, but the display is limited to a first-in, first-out scrolling display of fixed lines.
It is not possible to capture the pattern of a defective part two-dimensionally as a still image and analyze it in detail using digital image processing calculations. In addition, as seen in Japanese Patent Application Laid-Open No. 62-8045, two lines of one-dimensional image sensor cameras are arranged in the line running direction, and the first stage one-dimensional image sensor camera and defect detection circuit capture and detect defective parts. An example of capturing the pattern of a defective part as a two-dimensional still image and displaying it on a monitor using the second-stage one-dimensional image sensor camera, an A/D conversion circuit that converts analog signals to digital, and an image storage circuit using the same timing. However, characteristic analysis of defective parts can only be performed using electronic circuits and simple processing. Further, in this example, when driving the second-stage one-dimensional image sensor camera, it does not function unless the speed of the inspection object travel line is reduced. Therefore, it cannot be applied to high-speed and continuous running lines. Furthermore, as mentioned above, two lines of one-dimensional image sensor cameras must be arranged in the line running direction, which increases the cost. In particular, when defects occur continuously, the ability to seamlessly capture multiple two-dimensional still images of the pattern of the defect has not yet been reached.

[Problems to be Solved by the Invention] Regarding a method for detecting images of defects in a continuously running sheet-like object, it is possible to detect images of defects using video signals output from a one-dimensional image sensor camera that captures the pattern of a sheet-like object. When performing this process, a series of one-dimensional image sensor cameras are arranged in the running direction of the line, making it possible to capture the pattern of the defective area as a two-dimensional still image at high speed and continuously running the line, and perform detailed analysis using digital image processing calculations. In particular, it provides a method for seamlessly capturing a plurality of patterns of defective parts as two-dimensional still images even when the defective parts occur continuously.

[Means for solving the problem] A light source that illuminates a continuously moving sheet-like object, a one-dimensional image sensor camera that captures the pattern of the sheet-like object in one dimension, and a video signal that is repeatedly output from the one-dimensional image sensor camera. In a method of detecting a defective part image using a primary defect detection circuit that detects a defective part and an image processing device connected thereto,
The image processing device includes an analog/digital converter (hereinafter referred to as A) that converts the video signal into digital image data.
/D converter), an image recording circuit having at least two first-in, first-out type image memories for recording digital image data for repeated video signals capable of representing a defective part as a two-dimensional image, and a digital image stored in the image memory. A solution to this problem is to use multiple storage image memories that can transfer data. The present invention is a method for detecting images of defects in a continuously running sheet-like object using an image processing device, characterized in that the following image recording circuit and a plurality of storage image memories are used in the image processing device. There is. That is, the image recording circuit is provided with at least two A/D converters with high-speed conversion performance capable of converting the video signal into digital image data and a first-in, first-out type image memory for recording the digital image data. It is assumed that the image memory has a storage capacity capable of representing a defective part as a two-dimensional image. When the primary defect detection circuit detects a defect, one of the two first-in, first-out image memories is used to record still image data of the defect pattern with the defect in the center, and when the recording is completed, a plurality of separately prepared image memories are recorded. transfer the data to one of the image memories for storage,
Based on the contents of this storage image memory, the pattern of the defective part is analyzed in detail by digital image processing calculations. If defective parts occur continuously, the defective part pattern still image data is recorded in the other of the two first-in, first-out image memories so that the defective part is located in the center. By the time this recording is completed, the first-in-first-out image memory side has completed transferring data to one of the plurality of storage image memories prepared separately. By alternately using the shape image memories, it is possible to capture a plurality of defect pattern still images seamlessly. In this way, a still image of the defect pattern can be captured with one series of one-dimensional image sensor cameras without reducing the traveling speed of the sheet-like object.
It is also possible to perform detailed analysis of defect patterns using digital image processing calculations, making it possible to deal with cases where defective parts occur continuously.

[Embodiment] An embodiment of the present invention will be explained based on the drawings. FIG. 1 shows the configuration of an apparatus implementing the defect image detection method of the present invention. In the figure, 1 is a continuously running sheet-like nonwoven fabric. A vehicle with a width of approximately 350 mm was run at approximately 100 m/min. 2 is a light source, and 3 is a one-dimensional image sensor camera.The non-woven fabric 1 is sandwiched between the light source 2 at the bottom and the one-dimensional image sensor camera 3 at the top. did. The light source 2 may be a straight-tube high-frequency fluorescent lamp 2a that can irradiate the entire cloth width of about 350 mm. In this example, the driving frequency is 30kHz, 2
It was set to 0W. The one-dimensional image sensor camera 3 uses a one-dimensional solid-state image sensor, captures the pattern of the nonwoven fabric 1 in one dimension, and repeatedly outputs it as a video signal Vi. In addition, this repetitive scanning pulse signal S for repetitive timing
T is also output at the same time. 4 is a primary defect detection circuit which receives the video signal Vi and the repetitive scanning pulse signal ST, performs a rough inspection to detect defects by analog binarization processing, etc. within the inspection range of the nonwoven fabric 1, and detects defects. Detection signal Po
Output. Reference numeral 5 denotes an image processing device, which has an image recording circuit 5a and a plurality of storage image memories in addition to general image processing computer functions. The image recording circuit 5a
prepares an A/D converter 5a3 with high-speed conversion performance capable of digitally converting the video signal Vi, and furthermore,
First-in, first-out image memories 5a1 and 5a2 store digital image data obtained from the converter 5a3.
Prepare at least two pieces as shown in the figure. In this embodiment, the plurality of storage image memories are 1 byte/pixel data, 1024×102, as shown in 5b1 to 5b8.
Eight frames of 4 pixels/frame and approximately 1MB were prepared. The configuration of the one-dimensional image sensor camera 3 is shown in FIG. The one-dimensional image sensor camera 3 includes, for example, an imaging range 3 in a direction perpendicular to the traveling direction of the nonwoven fabric 1 described above.
Lens 3a and light receiving element 102 for imaging at 50mm
It is equipped with a one-dimensional solid-state image sensor 3b composed of four pixels, and passes through an electronic scanning circuit 3c that electronically scans photoelectric conversion signals for 1024 pixels obtained by the one-dimensional solid-state image sensor 3b every 0.125 ms to a sampling and hold circuit 3d. by 1
The photoelectric conversion signal of each pixel is converted into an analog continuous signal at a clock cycle of 2 MHz and outputted as a video signal Vi. In addition,
The one-dimensional image sensor camera 3 also simultaneously outputs a repetitive scanning pulse signal ST for video signal repetition timing necessary for each process of the primary defect detection circuit 4 and the image processing device 5. FIG. 3 shows the operation signal waveform of the one-dimensional image sensor camera 3. (a) shows a repetitive scanning pulse ST which is a repetitive pulse of 0.125 ms, and (b) shows a signal waveform of a video signal Vi capturing the pattern of the nonwoven fabric 1 for each electronic scan. Note that the horizontal axis indicates time t. Repetitive scanning pulse S
T is set to a DC level of 0 to 5 V, and the video signal Vi is set to a DC level of approximately 0 to +2 V depending on the size of the aperture of the lens 3a. A defective portion exhibiting light attenuation characteristics in the case of the transmission method appears on the video signal Vi as a waveform Si in which the voltage value partially decreases. The configuration of the primary defect detection circuit 4 is shown in FIG. The video signal Vi and the comparison signal Vt set by the comparison signal setting section 4b are input to the comparison section 4a, and the portion of the video signal Vi smaller than the comparison signal Vt is pulsed in order to capture the defect waveform Si exhibiting the dimming characteristic. Output. In addition, in order to operate within the original inspection range of the nonwoven fabric 1, an inspection range limiting section 4c is provided, and the inspection range is about 3
The limit pulse corresponding to 00 mm is used as a reference as the repetitive scanning pulse signal ST. That is, if the binarized pulse captured by the comparator 4a and the inspection range limiting pulse generated by the inspection range limiter 4c are ANDed in the arithmetic unit 4d, the defect waveform Si exhibiting the light attenuation characteristic within the original inspection range is obtained. A defect detection signal Po corresponding to the defect detection signal Po can be output. Primary defect detection circuit 4
The operating signal waveform of is shown in FIG. (a) is the video signal Vi
and the comparison signal Vt. (b) shows the output signal of the comparator 4a. The defect waveform Si showing the light attenuation characteristic is the comparison signal Vt.
Since it is smaller than Pi, it is output as a pulse as shown in (b). Further, a portion where the level of the video signal Vi is low outside the original inspection range of the nonwoven fabric 1 is similarly outputted as Pb shown in (b). Here, as described above, based on the repetitive scanning pulse ST shown in (c), an inspection range limiting pulse corresponding to about 300 mm of the nonwoven fabric 1 shown in (d) having a certain delay time with respect to (c) Pw is generated, and the output signal of the comparator 4a shown in (b) and the inspection range limiting pulse Pw shown in (d) are generated.
By performing an AND operation in the calculation section 4d, it is possible to output a defect detection signal Po corresponding to the defect waveform Si that shows the light attenuation characteristic within the original inspection range, as shown in (e). Note that the horizontal axis indicates time t. The configuration of an image recording circuit 5a as a component of the image processing device 5 is shown in FIG. Video signal Vi
is input to the A/D converter 5a3, and the repetitive scanning pulse ST
, the defect detection signal Po is input to the timing control section 5a5. Digital image data obtained from the A/D converter 5a3 is recorded in first-in, first-out image memories 5a1 and 5a2 under the control of the board controller 5a6. As described above, the first-in first-out image memories 5a1 and 5a2 have a capacity of about 1 Mbyte and can record image data of 1024×1024 pixels at 1 byte/pixel data. Note that the A/D converter 5a3 has a digital conversion speed of 12
×106 times/s, and the first-in first-out image memory 5a
1, 5a2 with a performance that can follow the digital conversion speed described above, the repetition period output from the one-dimensional image sensor camera 3 is 0.125 m.
The digital image data of 1024 pixels corresponding to the video signal Vi every s is stored in the first-in first-out image memory 5a1,
5a2 can be recorded sequentially. The digital image data whose recording and accumulation have been fixed are transferred to a separate image memory for storage via a data-only bus 5c via a data-only bus interface 5a4. In addition, the CP in the image processing device
Various control parameters etc. from U are sent to board control section 5a6 via control bus 5d and control bus interface 5a7. 7 and 8, the video signal V
The operation of recording digital image data obtained by A/D converting i into the first-in, first-out image memories 5a1 and 5a2 is shown. FIG. 7 shows normal operation without a defect, and FIG. 8 shows operation when a defect occurs. Note that the horizontal axis indicates time t. (a) shows the defect detection signal Po, (b) shows the repetitive scanning pulse signal ST, (c) shows the video signal Vi, and (d) shows the video signal Vi converted into 1 byte/image data by the A/D converter. Shows the pattern converted to digital image data. In addition, (e) and (f) are first-in first-out image memories 5a1 and 5.
The data structure of a2 is shown. The normal operation without defective parts will be explained with reference to FIG. Since no defective portion has occurred, there is no defect detection signal Po. After the repetitive scanning pulse signal ST, the video signal Vi is repeated for each scan. The A/D converter 5a3 converts the video signal V based on the repetitive scanning pulse signal ST input to the timing controller 5a5.
i is divided into 1024 pixels per scan, A/D converted, and this is repeated for each scan, and is recorded as digital image data at the head of the matrix of the first-in, first-out type image memory 5a1. That is, as shown in (e), one row corresponds to one scan of the video signal Vi. Since the first input digital image data is shifted sequentially, the first-in first-out image memory 5a1 is always filled with digital image data of 1024 x 1024 pixels, as shown in the shaded area in (e), and the old digital image is deleted every scan. Data is discharged from the end of the matrix for one scan, and a new one
The double-scanning digital image data is written at the beginning. Therefore, as shown in (d), one scan worth of digital image data N for the current video signal Vi
o. 1 is recorded in the 0th row and 0th column to the 1023rd column as shown in (e), and the digital image data No. 1 for one past scan is recorded as shown in (d). 511 is recorded in row 511, column 0 to column 1023, as shown in (e), and digital image data No. 511 for one past scan is recorded as shown in (d). 1023 is (
As shown in e), it is recorded in the last 1023rd row, 0th column to 1023rd column. Note that the other first-in, first-out image memory 5a
2 has exactly the same operating principle, but when there are no defective parts, there is no need to write digital image data. The operation when a defective portion occurs will be explained with reference to FIG. The same operation as shown in FIG. 7 is repeated until a defective part is detected, but when a defective part occurs, the defect detection signal P is
o is output, and from this, the first-in, first-out image memory 5a
The first-in, first-out operation is stopped after a delay time Td corresponding to the number of scans until half of the capacity of the image memory 5a1 is filled, and the digital image data in the first-in, first-out type image memory 5a1 is fixed. Thereby, it is possible to arrange the data such that the defective part is located in the center of the image. In this embodiment, the number of scans until half of the capacity of the first-out image memory 5a1 is filled is approximately 512 scans, and the delay time Td is 0.125×512=64 ms per 0.125 ms/scan. Incidentally, even if the next defective portion occurs within the delay time Td and the defect detection signal Po is input to the timing control section 5a5, it will not be accepted for the digital image data fixing operation. Also, the other first-in first-out image memory 5a
2 is one of the above-mentioned first-in, first-out image memories 5a1.
From the time when the digital image data is fixed, a first-in first-out data writing operation similar to that shown in FIG. 7 is started to prepare for the next defective portion. FIG. 9 shows the operation when defective parts occur continuously. (a) is a view from above of the nonwoven fabric 1 that is continuously running at high speed, and the imaging section of the one-dimensional image sensor camera 3 is point ■, and point ■ is the part before entering the imaging section, and points ■ and point ■ The time interval indicates the delay time Td. Point (2) is a region past the imaging section, and the time interval between points (2) and (2) also indicates the delay time Td. The ×, ●, and ○ marks indicate defective parts A, B, and C, and it is assumed that A, B, and C run continuously in that order. The defective portion A marked with a cross and the defective portion B marked with a black circle are continuous defects that run approximately at the time interval of the delay time Td. Defect part B marked with ● and defect part C marked ○
is a continuous defect that runs at a time interval approximately twice the delay time Td. In the horizontal direction of the figure, (A) to (C) and (D) to (G) are approximately the case where the delay time Td has elapsed, and (C) and (D) are the minute time ΔT elapsed (ΔT<< The flow in the case of Td) is shown. (b) is a diagram of the image data format of the first-out type image memory 5a1 corresponding to (a), and (c) is a diagram of the image data format of the first-out type image memory 5a2 corresponding to (a). (stomach)
When moving from the point in time to (b), the defective part A marked with an x enters the imaging section of the one-dimensional image sensor camera 3, the primary defect detection circuit 4 captures the defective part A and outputs the defect detection signal Po, and the timing control section 5a5 receives this. From this point, the first-in, first-out operation is stopped after a delay time Td corresponding to the number of scans until half of the capacity of the first-in, first-out image memory 5a1 is filled, and the digital image data in the first-in, first-out image memory 5a1 is fixed. c), which is the state shown in FIG. 8(e). Here, the digital image data in the first-in, first-out type image memory 5a1 whose recording has been fixed is started to be transferred to the storage image memory 5b1. Further, the other first-in first-out image memory 5a2
At the time when the digital image data in the first-in, first-out image memory 5a1 is fixed, a first-in, first-out data writing operation is started to prepare for the occurrence of the next defective part. (
First-in first-out image memory 5a1 captured at point c)
The digital image data can be imaged as a two-dimensional still image just for the time period Td before and after the defective part A marked with an x. (d) is even more △ than at (c)
This is the timing when time T has elapsed, and if the defective part B marked with ● has entered the imaging section of the one-dimensional image sensor camera 3, then from this point on, the first-in first-out image memory 5a2
The first-in first-out operation is stopped after a delay time Td until half of the capacity of the first-in first-out image memory 5a is filled.
The state in which the digital image data of 2 is fixed is (c)-
(e). At this point, the data transfer performance is set such that the transfer of the digital image data in the recording-fixed first-in-first-out image memory 5a1 to the storage image memory 5b1 is completed, and this time, the recording-fixed first-in, first-out image memory 5a2 is transferred. Transfer of the digital image data to the storage image memory 5b2 is started. moreover,
The first-in, first-out image memory 5a1 that has completed the transfer to the storage image memory 5b1 starts a first-in, first-out data writing operation from the time when the digital image data in the first-in, first-out image memory 5a2 is fixed to prepare for the next defective part. . Below (F) and (G) are the same diagrams as (B) and (C). In this way, by alternately using two first-in, first-out image memories and transferring data to multiple storage image memories, multiple still images of defect patterns can be seamlessly captured for continuous defects. be able to. The configuration of the image processing device 5 is shown in FIG. The storage image memories 5b1 to 5b8 are comprised of a data-only bus interface 5b11, a memory section 5b12, and a control bus interface 5b13. When storing digital image data in the memory section 5b12, The data is stored from the data-only bus 5c via the data-only bus interface 5b11. Further, 5g is a CPU and a main memory, and various control parameters from the CPU to the memory 5b12 are transmitted from the control bus 5d to the control bus interface 5b13.
sent via. 5e is image memory for calculation, 5f is image processing processor, CPU and main memory 5g
By the above program, storage image memory 5b1 to 5
The digital image data of b8 is stored in the image memory 5e for calculation.
By moving to the image processing processor 5f and performing digital image processing calculations, detailed analysis of the defective portion becomes possible. In this embodiment, the calculation image memory 5e has a capacity of 1 byte/
Pixel data, 1024 x 1024 pixels/frame, approx. 1
I prepared 8 frames of M bytes. As described above, the one-dimensional image sensor camera 3 and the primary defect detection circuit 4
, the conversion of the pattern of the nonwoven fabric 1 into digital image data by the image recording circuit 5a and the like and the image processing operation of the digital image data use completely different image memories.
Detailed analysis of the defective part by image processing calculations can be executed sequentially under the control of the CPU after the defective part is detected. 5h
is a fixed magnetic disk memory, and 5i is a floppy disk, which is used to store programs and data. 5j is a monitor, 5k is a printer, 5m is an image hard copy,
It is used to display the results of the detailed analysis of the defective area, print it out on print paper, or make a hard copy of the image of the defective area. As described above, the high-speed conversion performance of the A/D converter 5a3 that converts the video signal Vi into digital image data, and the first-in, first-out image memories 5a1 and 5a2
High-speed image data writing performance from first-in, first-out image memories 5a1 and 5a2 to storage image memory 5b
1. Utilizing the high-speed image data transfer performance to 5b8,
Two first-in, first-out image memories 5a1 and 5a2
A plurality of storage image memories 5b1 are used alternately.
, 5b8, the defective pattern of the nonwoven fabric 1 can be captured as a still image while continuously running at high speed without temporarily decelerating the running speed of the nonwoven fabric 1, and can be analyzed in detail using digital image processing calculations. . Also,
Even when defects occur continuously, a plurality of defective patterns of the nonwoven fabric 1 can be seamlessly captured as still images, and detailed analysis can be performed by digital image processing.

[Effects of the Invention] The method for detecting images of defects in a traveling sheet-like body according to the present invention detects the defect pattern of a sheet-like body while continuously traveling at high speed without temporarily decelerating the traveling speed of the sheet-like body. This system captures images as still images and enables detailed analysis using digital image processing calculations.Furthermore, it has the feature of being able to respond as described above even when defects occur continuously. As a result, it is now possible to perform detailed analysis of defects in a sheet-like object that is traveling continuously at high speeds, which was previously impossible. It is possible to identify the type of parts and grade the sheet-like objects, making it possible to greatly improve the efficiency of the work of rewinding the sheet-like objects. Furthermore, since the type of defective part and the extent of the defect can be immediately determined on the product production line, process feedback to the product production line can also be provided immediately. Furthermore, since there is a limit to the number of storage image memories in the image processing device, it is not possible to record images of an infinite number of consecutive defective parts; This can be handled by outputting a feedback conclusion. In this embodiment, one one-dimensional image sensor camera is used.
However, it is possible to handle the case of multiple one-dimensional image sensor camera systems by expanding the number of each element, and defective parts can be captured by separate cameras at the same timing in the traveling direction. The defect pattern can also be detected as a still image.

[Brief explanation of the drawing]

FIG. 1 is a configuration explanatory diagram in an embodiment of the present invention.

FIG. 2 is a configuration explanatory diagram of a one-dimensional image sensor camera in an embodiment of the present invention.

FIG. 3 is an operation signal waveform diagram of the one-dimensional image sensor camera according to the embodiment of the present invention, in which (a) shows a repetitive scanning pulse signal and (b) shows a waveform of a video signal.

FIG. 4 is a configuration explanatory diagram of a primary defect detection circuit in an embodiment of the present invention.

FIG. 5 is an operation signal waveform diagram of the primary defect detection circuit in the embodiment of the present invention, in which (a) is a video signal and a comparison signal (b) is a comparison section output (c) is a repetitive scanning pulse signal (d). The inspection range limit output (e) shows the waveform of the defect detection signal.

FIG. 6 is an explanatory diagram of the configuration of an image recording circuit in an image processing apparatus in an embodiment of the present invention.

FIG. 7 is an operation signal waveform diagram of the image recording circuit when no defective portion occurs in the embodiment of the present invention.

FIG. 8 is an operation signal waveform diagram of the image recording circuit when a defect occurs in the embodiment of the present invention, in both FIGS. 7 and 8, (a) is a defect detection signal (b) is a repetitive The return scanning pulse signal (c) is a video signal waveform (d) is a representation of the image data (e) is the representation of the image data in one first-in, first-out image memory (f) is the image data in the other first-in, first-out image memory FIG.

FIG. 9 is an explanatory diagram of image recording operation when defective parts occur continuously in the embodiment of the present invention; (a) is a diagram of the nonwoven fabric viewed from the one-dimensional image sensor camera side; (b) is one of the nonwoven fabrics; Diagram (c) of the image data in the first-in, first-out image memory shows the image data in the other first-in, first-out image memory.

FIG. 10 is an explanatory diagram of the configuration of a storage image memory and other components in the image processing apparatus in the embodiment of the present invention.

[Explanation of symbols]

1. Continuously running sheet material (non-woven fabric) 2. Light source 3 One-dimensional image sensor camera 4 Primary defect detection circuit 5 Image processing device 5a Image recording circuit

Claims (1)

[Claims] Claim 1: A light source that illuminates a continuously moving sheet-like object, a one-dimensional image sensor camera that captures the pattern of the sheet-like object in one dimension, and a defective part detected from a video signal repeatedly output from the one-dimensional image sensor camera. In a method for detecting a defect image of a continuously running sheet-like object using a primary defect detection circuit and an image processing device, the image processing device includes an analog/digital converter for converting the video signal into digital image data, and a defect portion. at least two first-in, first-out image memories for recording the digital image data for repeated video signals capable of representing a two-dimensional image;
1. A method for detecting images of defects in a continuously running sheet-like object, the method comprising using an individual image recording circuit and a plurality of storage image memories capable of transferring digital image data stored in the image memories.