FR2785625A1 - System to identify and locate faults in moving woven fabrics has cameras where the image signals are processed to show a probable fault by type and confirm the finding as a fabric fault automatically - Google Patents

Abstract

The woven fabric inspection system has at least one camera (416A-416X) to record an image of part of the unwinding fabric, and transmit image signals. The image signals are processed (604-606) so that, initially, they are analyzed for fault detection and a detected fault is validated by at least one further image signal for confirmation of the located fault. The image processing system (604-606) is structured so that the image of the first signal and the image of each of the second signals give an occupied intersection as the location of a probable fault. The second image signal is processed with the same criteria as the first signal processing, so that a probable fault shown in the first image is confirmed as a fault by the second image. The image processing system (604-606) has an initial stage, where the nature of the fault is identified from a number of possible conditions, to give an evaluation of the probable presence of a fault. Each of the second stages operates with confirmation criteria based on the nature of the fault determined in the first stage.

Description

 The present invention relates to a method and a device for inspecting fabric during weaving.

 The fabric inspection devices known to date use a linear camera which continuously captures an image of the fabric, and an image processing means which processes the images coming from the linear camera to detect weaving defects. These devices are very expensive and cannot be made profitable.

 The present invention intends to remedy these drawbacks.

To this end, the present invention relates, according to a first aspect, to a fabric inspection device, characterized in that it comprises:
at least two cameras adapted to each capture an image of part of the tissue and to transmit a signal representative of the image captured,
a suitable selection means, for each camera, for selecting an image part, and
a processing means adapted to process said image parts and to detect therein at least one weaving defect.

Thanks to these arrangements, an image part can be chosen for each camera which corresponds:
-or to the part of the image of better quality (due to imperfections of the lens or lighting, for example,
-or to the part of the image which concerns an area of the tissue which has not yet been inspected,
- either to a part of the image which is most prone to present defects.

 By thus selecting parts of images, the present invention avoids the processing means from processing all of the captured images.

 According to particular characteristics, the optical fields of the cameras extend jointly over the entire width of the fabric to be inspected.

 According to other particular characteristics, the selection means is adapted to select parts of images which, jointly extend over the entire width of the fabric to be inspected.

 Thanks to each of these arrangements, the entire fabric can be inspected, during the deptocation of this fabric or of the cameras, facing the fabric.

According to other particular characteristics, the inspection device as succinctly described above comprises:
at least one camera adapted to capture an image of the last thread inserted into the fabric, and to provide a signal representative of said image, and
an image processing means adapted to process signals representative of images captured by each camera, and to detect in said image, the image of at least one weaving defect.

Thanks to these arrangements, as soon as the thread has been inserted into the fabric, its image is captured by the camera and processed by the processing means. A fault is therefore immediately detected, which allows:
-to stop the weaving machine as soon as a defect has been detected,
-reducing the amount of fabric that is woven after the appearance of the defect and,
-if the weaving machine allows, loosen the fabric and repair the defect.

According to other particular characteristics, the inspection device as briefly described above is associated with a weaving machine and comprises:
at least one camera adapted to capture an image of the tissue, during a shooting period, and to emit a signal representative of the captured image,
a synchronization means adapted to emit a signal representative of the position of the comb of the weaving machine, and
a processing means adapted, for each camera, to process a signal representative of the image leaving said camera and corresponding to a shooting period during which the comb does not strike the fabric.

 Thanks to these provisions, during the shooting period, the fabric is practically immobile and the processed image comprises little or no motion blur. The processing means then makes it possible to detect more defects than in the case where the shooting period does not exclude the period of striking the comb on the fabric.

 According to other particular characteristics, the processing means is adapted to process a signal representative of image leaving one of the cameras more frequently than a signal representative of image leaving another camera.

 Thanks to these provisions, an area of an object to be inspected, for example, a fabric or a manufactured object, in which it is considered that several shots are necessary to properly detect a defect, or because in this area the faults are more difficult to detect, either because in this zone the faults are more likely to appear, or either, because in this zone, the object presents a faster movement is less well lit than with regard to another camera is the object of image processing more frequent than other areas of said object.

 According to other particular characteristics, the inspection device as briefly described above comprises a means for estimating the probable position of at least one defect on said tissue, and the treatment means is adapted to more frequently treat a image representative signal leaving each camera which corresponds to a so-called probable fault position than a signal from another camera.

 Thanks to these provisions, the positions where it is likely to find a defect are inspected more frequently than the positions where it is unlikely to find a defect. The treatment of the defect, for example the stopping of the weaving machine, the untwisting, the cleaning can then be done more quickly after the appearance of the defect.

 In addition, in the case where an exhaustive inspection of the tissue is not possible, for example because the image processing applied by the processing means to the signals from the cameras is very complex, the provisions of the present invention allow a exhaustive inspection of positions where it is deemed likely to find a defect.

According to a second aspect, the present invention relates to a process for inspecting tissue, characterized in that:
it uses at least two cameras adapted to each capture an image of part of the tissue and to transmit a signal representative of the captured image,
it comprises a selection operation during which, for each camera, an image part is selected, and
a processing operation during which said image parts are processed and at least one weaving defect is detected there.

The present invention also relates to:
a weaving machine, characterized in that it includes an inspection device as briefly described above,
-a tissue cutting machine, characterized in that it comprises an inspection device as succinctly set out above.

 The present invention also relates to a memory which stores instructions of a program adapted to be executed by a processor, and allowing the implementation of the method of the present invention.

 The advantages of this process, this memory and this weaving machine being the same as those of the first aspect of the invention, they are not repeated here.

The invention will be better understood on reading the description which follows, made with reference to the appended drawings in which:
FIG. 1 represents, in side view, a fault detection device according to the present invention;
FIG. 2 represents optical fields of cameras incorporated in the device illustrated in FIG. 1.

FIG. 3 represents, diagrammatically, an electronic microprocessor circuit intended for processing images, incorporated in the fault detection device illustrated in FIG. 1; and
FIG. 4 represents an operating flow diagram of the circuit illustrated in FIG. 3.

 In FIG. 1, are represented: a beam unwinder 101, schematically represented in the form of a spool of threads in the course of unwinding, a fabric reel 102, schematically represented in the form of a spool of fabric in winding course, twenty-four cameras 116A to 116X (the last twenty-three being, in side view, masked by the first), each provided with a lens 126, and all substantially oriented towards the area of striking a comb 104 of a weaving machine (not shown), a light source 103 adapted to illuminate the strip of fabric in the vicinity of the optical axis of the cameras 116A to 116X, on the same side of the fabric as these cameras, a blade 105, a blade 106, a motor for setting in motion the comb 104 and the blades 105 and 106, which emits, cyclically, a pulse, during the movement of these elements of the weaving machine, computer 114, -a computer screen 115 connected to computer 114, and -a link network sound 113.

 The thread unwinder 101 and the fabric winder 102 are part of a weaving machine of a type known in the textile industry.

 The cameras 116A to 116X, are each made up of a known type of matrix camera. It has, for example, a monochrome matrix sensor, that is to say all of whose elementary photosensitive surfaces (known as "pixels") are on a rectangular table formed by rows and columns. Each camera lens 126 is of known type. It is suitable for focusing on the photosensitive surface of the sensor of the camera with which it is associated, a sharp image of a rectangular area which surrounds the striking area of the comb. The optical fields of cameras 116A to 116X overlap laterally (see Figure 2). The cameras are preferably synchronized with one another according to known techniques. They take shots simultaneously and simultaneously supply electrical signals representative of the quantities of light perceived during a shot by all the photosensitive points of their sensor.

 The light source 104 is of known type. It is, for example, made up of a neon tube, supplied by an electrical supply supplying a high frequency current, said supply being itself supplied by an alternative electrical supply (or, according to a preferred variant, continuous).

 The computer system, comprising in particular the computer screen 115 and the computer 114, is zero-rated with regard to FIG. 3. Finally, the network link 113 allows the computer 114 to communicate with a computer network, for production management.

 In FIG. 2, parallel threads 201 and 202 are shown intended to form the warp threads of a fabric (here a canvas), after a comb 104 has struck a weft thread 208 in the fabric already formed. In FIG. 2, the comb 104 is shown when the wire 208 is struck.

 The cameras 116A to 116X observe an optical field 204.205, 206, ..., in which the set of successive lines 209, 210.221, .... relates to the last weft thread 208 which has been struck by the comb 104.

 The optical fields of the cameras partially overlap in the direction perpendicular to the direction of movement of the fabric being woven.

 It is observed that in the optical field of the camera 116B, there is a suspect wire 207. This wire is considered to be suspect, either because it has already been the subject of faults previously detected, or because, without being defective, he has a higher abnormality than the other sons. For example this thread is very tight, very thin, ...

 In FIG. 3 are represented, in a computer system 30, linked together by a data and address bus 302, a central processing unit 306 consisting of a processor, a read-only memory 305, a random access memory 304 and a port input / output 303.

 These elements are well known to those skilled in the art of computer systems and are therefore not more detailed here.

I / O port 303 is connected to
a keyboard 301, to receive instructions therefrom from the user,
the network link 113,
each camera 116A to 116X, by means of a multiplexing and digitization card 307, which successively supplies a digital signal representative of the image captured by the sensor of each camera,
-screen 115,
an input 309 for synchronization signals of the comb, on which a signal coming from the weaving machine and representative of the progress of at least one phase of the operating cycle of said weaving machine, is present, also called cyclic signal input from engine 107, and
an outlet 315 to an electrical cabinet for controlling the weaving machine and in particular for stopping the weaving machine.

 The read-only memory 305 stores, in a register program1, the operating program of the computer system 30 and, in particular, the fabric inspection program illustrated in the form of a flowchart in FIG. 4.

 The random access memory 304 stores in registers (i and data), variables, intermediate processing results and processing results of the operating program of the computer system 30.

 The central unit 306 is suitable for implementing the program, part of which is shown in FIG. 4.

In FIG. 4, we observe, first of all, an initialization phase 401 known to those skilled in the art of computer-based image processing systems. During this phase, from data stored in memory or provided by the user, via the keyboard 301, the central unit 306 performs the determination of thresholds, the width, the configuration of the digitization card. 307, its gain, its offset to the origin (known from
A person skilled in the art under the English name of offset), and the sequencing of cameras.

 The sequencing of the cameras is the order in which the signals transmitted by the cameras are analyzed by the central unit 306.

Preferably, the cameras 116A to 116X being numbered from 1 to 24, the sequencing begins, following the positioning of the wire by the comb 104, -by the two extreme cameras, numbered 1 and 24, then -by eleven of the intermediate cameras, for example, 13,3,17,7,21,11,5,19,9, 23,15, then - again, by the two extreme cameras numbered 1 and 24, then - by the eleven cameras remaining intermediates 14,4,16,6,20,10,18,8,22,12, 2.

 However, this sequencing is modified so: when a defect is detected in the optical field of one of the cameras: - immediately after, a new image taken by said camera is analyzed, in order to confirm or deny the presence of '' a defect, - the camera concerned is considered as observing a suspicious area.

 The sequencing then becomes: -by the two extreme cameras, numbered 1 and 24, then -the camera which observes the most suspicious area, -by eleven of the intermediate cameras, for example, 13,3,17,7,21,11 , 5,19,9, 23,15, with the exception, if necessary, of the camera which observes the most suspect area, then, again, by the two extreme cameras numbered 1 and 24, then again by the camera which observes the most suspect area, -by the eleven remaining intermediate cameras 14,4,16,6,20,10,18,8,22,12, 2 with the exception, if necessary, of the camera which observes the most suspicious area.

 According to a variant, two cameras are considered to be suspect and are inserted into the sequencing in place of the only camera considered to be suspect in the example above: -the last camera in the field of which a fault has been detected and confirmed after analysis of the second image, the last camera in the field of which a defect has been detected but has not been confirmed after analysis of the second image.

 More generally, the central processing unit 306 is adapted to process a signal representative of image leaving one of the cameras more frequently than a signal representative of image leaving another camera.

 To this end, in the particular embodiment of the invention illustrated in the figures, the central unit 306 performs an operation of estimating the probable position of at least one defect on said fabric, more frequently processes a signal representative of image coming out of each camera which corresponds to a so-called probable fault position than a signal from another camera.

 In the sequence presented above, and which is preferable when it is feared, on a given weaving machine, fabric edge defects, the central unit 306 considers that the edges of the fabric are likely positions of defects.

 The central unit 306 then performs a sequencing of the cameras to organize an order of cameras whose images are successively processed by it.

 As we have seen, the central unit 306 also considers that the position where the processing means has detected a fault which has caused the machine to stop is a probable fault position.

 Returning to the description of FIG. 4, then, the central unit 306 determines, during a test 402, whether the cyclic signal coming from the motor 107 has been received or not. This cyclic signal is representative of a position of the comb in the operating cycle of the weaving machine.

 When the result of test 402 is negative, it is repeated.

 When the result of the test 402 is positive, the central processing unit 306 performs, during an operation 403, a time delay, that is to say it waits for a predetermined duration to elapse. This duration makes it possible to wait until the comb is stopped or, at least, until the movement of the tissue induced by the striking of the comb is completed. During the same operation 403, the intermediate variable i, which represents the rank in the sequence, is initialized to the value 1.

 During an operation 404, the central unit 306 stores the images simultaneously from all the cameras 116A to 116X.

Then, during an operation 405, part of the image from the camera referenced to the i-th row of! he sequence determined during phase 401 is processed according to an algorithm which depends on its position: the extreme cameras referenced 116A or 116X, or even 1 and 24, give rise to image processing intended to detect edge defects ( correct width, thread retracted, thread breakage, thread protruding, edge too retracted, ...) while the other cameras are subject to another image processing (oriented, warp or weft defects and defects local).

 These image processing are well known to those skilled in the art and are therefore not detailed here. As an example, patent application FR-A-2 701 766 (Institut Textile de France, August 26, 1994) details some of these image processing algorithms.

 It is observed here that it is not the whole of each image which is processed during the operation 405 but only the central band, excluding the upper and lower bands which correspond to a part of the fabric whose image is processed during another operation 405. The height of the strip of fabric treated corresponds, in fact to approximately the length of fabric woven between two successive image treatments applied to two images originating from the same camera.

 Thus, the central unit makes a selection, for each camera, of an image part, and only processes said image parts to detect therein at least one weaving defect.

 In the embodiment described and shown, the optical fields of the cameras extend, jointly over the entire width of the fabric to be inspected (see FIG. 2) and the selection of parts of images carried out by the central unit selects parts of images which jointly span the entire width of the fabric to be inspected.

 For each defect detected, its position in the image makes it possible to determine whether it has already been spotted in the analysis of another image (taken by another camera), and which pick it is at. The position, along two axes, of this fault, as well as the part of the image which surrounds it and the parameters of the detected fault, are stored in memory by the central processing unit 306.

 When a fault is detected, the sequencing of the cameras is modified so that images taken by the considered camera are analyzed twice as often as the images taken by the other cameras, except for extreme cameras and any other suspicious cameras, knowing that, preferably, the number of suspicious cameras is limited to one or two. At the same time, when a fault is detected, a new image is taken by the camera and put in memory. This new image is analyzed in the same way as the first image. If a fault is also detected in the second image, it is confirmed and its types, position and images are stored in memory. On the contrary, if the second image has no defect, the defect observed in the first image is not taken into account.

 During a test 406, the central processing unit 306 determines whether a defect of one of the predetermined types (breakage of wire, etc.) or of a predetermined importance (dimension greater than a limit value, contrast with the tissue greater than another limit value) was detected, and validated, during operation 405.

 When the result of test 406 is positive, the weaving machine is stopped, operation 407.

 Then, during a test 408, the central processing unit 306 determines whether or not the fault can be repaired by untwisting and tightening. When the result of test 408 is positive, the weaving machine performs the unscrewing and reweaving of the fabric to the position of the defect and the repair of the defect, during an operation 409.

 When the result of the test 408 is negative, the central unit 306 triggers an alarm (flashing light, sound signal or computer message on a computer network comprising the part of the image on which a fault has been detected), during a operation 410.

 Each time the weaving machine is restarted after an operation 410, the central unit 306 performs the operation 411, while keeping in memory the images which had been captured and memorized just before the machine stopped (operation 407). Then, after a few cycles each comprising the analysis of images of a whole sequence of cameras, the central processing unit 306 performs operations identical to operations 402 to 404, then an image processing consisting in comparing the luminances (levels of gray) warp threads before and after you stop the machine.

 Preferably, this comparison is only carried out on images coming from camera i, one of which caused the machine to stop. Alternatively, all cameras are considered.

 For each camera of which two images are compared, knowing that two very different colors can correspond to very close gray levels or luminances, the central unit 306 takes the average of the gray levels along the warp wires, outside the area where a fault had been detected.

 It is known that, except in the event of a defect in positioning the optical axis of the camera lens perpendicular to the plane of the area of the tissue inspected, the warp threads correspond to vertical columns in the raster images. Also, the averages compared here are those of the gray levels on each vertical column of photosensitive dots, between the two images taken by the same camera (except in the zone where the defect has been detected, which is excluded from the average, so that the default does not distort the average considered). So, when the operator replaces a warp thread but gets the wrong color, the device detects this error and causes a new one! stopping the machine by keeping in memory not the most recent image but that which had preceded the first of the two stopping of the weaving machine.

 Following this search for a fault in the incorrect positioning of the chain wire, the operation then resumes at operation 405, with the same value of i.

 When the result of test 406 is negative, or following the operation 409, the variable i is incremented by a step equal to 1, during an operation 411.

 During a test 412, the central unit 306 determines whether the value of the variable i is equal to 14, or not. When the result of the test 412 is positive, an image taken by each of the cameras 1 and 24 is again memorized, operation 414 and the operation 405 is repeated.

 When the result of test 412 is negative, during a test 413, the central processing unit 306 determines whether the value of the variable i is strictly greater than 26, or not. When the result of test 413 is negative, operation 405 is repeated. When the result of test 413 is positive, test 402 is repeated.

 When the machine is finally stopped, at the end of the weaving of a roll of fabric, the central unit automatically publishes a visit sheet which contains the information concerning the detected faults which have not been repaired and, possibly , other faults detected. This visit form is transmitted over a computer network intended for production management.

 According to a variant not shown, the shooting intended for the storage of an image is carried out under the control of the central unit 306.

Operation 404 then comprises said triggering as soon as the tissue is stopped.

 Alternatively, the central unit successively considers the cameras in the optical field of which the most suspect wires are found, in descending order of suspicion.

 According to a variant, the device covered by the present invention, as set out above is dissociated from a weaving machine but associated with a fabric cutting table. To this end, the comb synchronization signals are replaced by a synchronization signal on the movement of the fabric under the cameras, for example from an encoder. In fabric cutting tables, like those of the French registered trademark LECTRA SYSTEMS, the cutting instruments are knurls, lasers or water jets.

 The implementation of the present invention then allows operators to be alerted of the presence of faults and to identify these faults and, in the case of automatic cutting tables, to the software for determining the positions of the parts to be cut, to order cutting the pieces of fabric outside the defective areas identified by the device.

Claims (10)

 1. Fabric inspection device, characterized in that it:  at least two cameras adapted to each capture an image of part of the tissue and to transmit a signal representative of the image captured,  a suitable selection means, for each camera, for selecting an image part, and  a processing means adapted to process said image parts and to detect therein at least one weaving defect.  2. Inspection device according to claim 1, characterized in that the optical fields of the cameras extend, jointly over the entire width of the fabric to be inspected.  3. Inspection device according to claim 2, characterized in that the selection means is adapted to select parts of images which, jointly extend over the entire width of the fabric to be inspected.  4. Inspection device according to any one of claims 1 to 3, characterized in that it comprises:  at least one camera adapted to capture an image of the last thread inserted in the fabric, and to provide a signal representative of said image, and  an image processing means adapted to process signals representative of images captured by each camera, and to detect in said image, the image of at least one weaving defect.  5. Inspection device according to any one of claims 1 to 4, characterized in that it is associated with a weaving machine and in that it comprises:  at least one camera adapted to capture an image of the tissue, during a shooting period, and to emit a signal representative of the captured image,  a synchronization means adapted to emit a signal representative of the position of the comb of the weaving machine, and  a processing means adapted, for each camera, to process a signal representative of the image leaving said camera and corresponding to a shooting period during which the comb does not strike the fabric.  6. Inspection device according to any one of claims 1 to 5, characterized in that the processing means is adapted to process more frequently a signal representative of image leaving one of the cameras than a signal representative of image from another camera.  7. Inspection device according to claim 6, characterized in that:  it comprises a means of estimating the probable position of at least one defect on said fabric, and  the processing means is adapted to process more frequently a signal representative of the image leaving each camera which corresponds to a so-called probable fault position than a signal coming from another camera.  8. A weaving machine, characterized in that it includes an inspection device according to any one of claims 1 to 7.  9. Fabric cutting machine, characterized in that it comprises an inspection device according to any one of claims to 7.  10. Fabric inspection process, characterized in that:  it employs at least two cameras adapted to each capture an image of part of the tissue and to transmit a signal representative of the captured image,  it includes a selection operation during which, for each camera, an image part is selected, and  a processing operation during which said image parts are processed and at least one weaving defect is detected there.