FR3129214A1 - Spreading equipment with fabric defect identification system and associated method - Google Patents

Abstract

The invention relates to a tissue spreading equipment comprising a system for identifying fabric defects comprising: - an image capture system on the crosshead adapted to capture an image in correspondence with a detection zone located downstream of the spreading strips and upstream of the spreading area; - a database in which fault definition characteristics are stored; - a processing system adapted to receive an image of the sheet from the image capture system, extrapolate characteristics of the image and carry out a comparison between the characteristics of the image with the characteristics defining defects; - an error signaling device functionally connected to the processing system. If the image characteristics match the defect defining characteristics, the image is classified as a defect and a control signal is generated for the error signaling device (Fig. 3)

Description

Spreading equipment with fabric defect identification system and associated method

Scope

In its most general aspect, the present invention relates to laying equipment for fabrics with a system for identifying defects when laying down such fabrics.

The present invention also relates to a method for identifying faults.

The present invention applies in particular, but not exclusively, to the textile field, and more specifically to the interior of automatic spreading machines, widely adopted in the mass production of elements for clothing, upholstery, linen, etc.

State of the art

As is well known in this specific technical sector, over the years there has been an increasingly important development of automatic equipment both with regard to the stage of spreading the sheets of fabric and in with regard to the next stage of cutting the various elements of the garments to be produced. The purpose of this automatic equipment is to reduce production operating times without losing the quality of the final product.

Although the automation currently carried out in the sector has already allowed a considerable development of the production process, this process still requires, in some cases, the human intervention of an operator, who must therefore remain close to the equipment, during different stages of spreading and cutting in order to be able to carry out operations, in particular but not only, of correcting anomalies.

In other words, during the various stages of production, an operator must often be present next to the machine to avoid errors upstream of production which would lead to a final product of unsatisfactory quality.

This clearly implies an unproductive use of the available manpower, both in terms of the operator time actually adopted in the daily cycle and in terms of the profit margin in relation to the costs incurred.

Nowadays, the main trend is therefore to develop automatic equipment that is increasingly autonomous in its basic functions, which makes it possible to use operator time more efficiently in tasks with a higher level of control.

Among the tasks still carried out in most cases by the operators is that of checking that the fabric which is laid out and from which the elements of clothing, padding, linen, etc. are then cut. to be produced is free from defects which would affect the final product. Examples of such defects can be related to yarn type, warp and weft direction, defects due to non-uniform dyeing, printing and finishing, to name a few. -ones.

Normally, in the fabric spreading equipment, the fabric can arrive in a roll and then be spread out to obtain a mattress made up of one or more sheets.

Instead of the rolled material, the spreading equipment can also spread the fabric that is already coming in layers.

In all cases, the fabric, whatever it is and whatever its "shape" of arrival, passes through the spreading equipment in order to obtain a mattress of individual sheets on top of each other, a fabric zig-zag (also called folded), a fabric that is zigzag and cut on one side or a fabric that is zigzag and cut on both sides.

On the other hand, for example, leather is not ironed with the spreading equipment, but still requires a manual spreading process, whereas when fabrics are used, the process is usually almost fully automatic.

In any case, as the fabric is spread, either manually or by means of automatic spreading equipment, the operator can see a part of the fabric and assess the presence or absence of defects.

In the event of defects observed, the operator can interrupt the spreading step to move on to a so-called "cut-recovery" step, a procedure necessary to "cover" the part containing the defect, in order to obtain an optimally spread sheet of tissue.

However, it is possible that the operator is distracted or not perfectly competent for the aforementioned work, thus not identifying certain faults.

In addition, the light may not be optimal to see the defects with the naked eye.

All of this has a negative impact on production times, as the visual identification of less obvious stains or defects can take longer and again leads to an inefficient use of the working time of qualified personnel.

In addition, defining the boundaries of a manually identified defect often presents gross errors with a consequent increase in costs related to the greater amount of material removed with the defect.

For this reason, other solutions have recently been proposed to also automate this fault identification step.

Patent application 102020000004072 of the same applicant describes equipment with an automatic system for identifying defects on a spreading surface.

Patent application IT 201800005606 A1 in the name of Caron Technology S.r.l. similarly describes a spreading machine on which is provided a system for identifying any imperfections in the fabric spread on a spreading surface in order to exclude them in the next cutting step.

Although these solutions are proposed as decisive solutions to the problem of automating the fault detection step, they have a series of drawbacks.

In particular, based on the proposed methods, once the presence of the defect has been detected, the sheet must be brought back with respect to the spreading direction so as to eliminate the part containing the defect, consequently causing a loss, also in this case, useful production time.

Furthermore, the distance of the defect identification system from the sheet requires a definition of the image such that it allows the detection of even the smallest defects, with a consequent increase in the costs of the instruments adopted.

In addition, the entire equipment creates an arrangement of light and dark areas on the spreading surface in such a way as to make it more difficult to identify the defect.

Furthermore, as current spreading and cutting equipment is designed to work on mats of stacked sheets, in the event that the fabrics have characteristics of even partial transparency, their overlapping can make it even more difficult to identify the defect.

For this reason, the technical problem underlying the present invention is that of solving the drawbacks of the state of the art.

Another purpose is to increase the rate of removal of a defect from a tissue.

Yet another object of the invention is to optimize the precision of the detection and elimination of a defect in a fabric.

Another goal is to provide equipment that can be used interchangeably for any type of fabric without the need for a dedicated configuration.

Another aim is to provide equipment that is simple to use and does not require complex specific training.

The idea of a solution underlying the present invention is to identify, during a continuous spreading step, the best times for the steps of detecting and eliminating a defect from a sheet of fabric, also by relation to the next cutting step.

By "defect" in the present development is meant a non-conformity of a part of fabric with the standard characteristics of the fabric itself, therefore not considering as a defect, for example, a weft of a fabric which is part of the very nature of a specific material used.

Based on this solution idea, the technical problem is solved by a spreading equipment for fabrics, comprising at least one movable frame, at least two spreading strips, at least one handling means for the spreading strips , a spreading structure comprising a spreading area for at least one sheet and at least one crosspiece, and a fabric defect identification system.

This system for identifying defects on fabrics comprises at least one image capture system positioned on the crosspiece oriented towards the spreading strips and adapted to capture at least one image of the sheet in correspondence with a detection zone located in downstream of the spreading strips and upstream of the spreading area. The defect identification system further comprises at least one database in which defect defining characteristics are stored and at least one processing system, active during the spreading of the sheet, connected to the at least an image capture system and linked to the at least one database.

The processing system is adapted to receive at least one image of the sheet from the image capture system, extrapolate features of the image, and perform a comparison between the features of the image with the defect defining features.

If the characteristics of the image match the defect defining characteristics, the image is classified as a defect and a control signal is generated for the error signaling device.

Advantageously, the equipment according to the invention makes it possible to identify a defect in the spreading phase, before the sheet reaches the spreading surface, thus being able to carry out the elimination immediately, without continuous advancement or recoil of the sheet. .

In addition, the specific position of the image capture system, very close to the sheet and always focused on a limited area of the sheet itself, allows maximizing the accuracy of defect detection.

In addition, the aforementioned position of the image capture system is optimal with regard to the illumination of the tissue.

Also, because the fabric is "falling" over the spreading surface, and therefore harnessing gravity, it is virtually unaffected by the slight ripples that often occur on the spreading surface, even more marked in the in the case of spreading out as a mattress one sheet on the other, these undulations being of course also able to mask or make a defect less obvious or, on the contrary, to create arrangements of the sheets of fabric which can be wrongly identified as defects.

Preferably, the equipment comprises an error signaling device operatively connected to the processing system, and the latter generates a control signal for the error signaling device in the event of a match of the characteristics of the image with the defect defining characteristics and the classification of the image as a defect.

Advantageously, there is therefore an automatic response from the equipment to the identification of the fault.

According to a preferred embodiment, the tissue spreading equipment instead comprises a plurality of image capture devices, arranged at a predetermined distance along the traverse towards the detection zone.

Advantageously, the present solution allows an additional subdivision into sectors of the detection zone with an optimal result on the precision of detection of defects.

According to another preferred embodiment, the image capture system comprises at least one linear camera, suitable for framing the entire detection zone.

Advantageously, the present solution makes it possible, with a single specific device, to frame a strip of fabric at a time, with a simplification of the system and with a result having the required precision.

Preferably, the equipment according to the invention further comprises a lighting device for the detection zone.

Advantageously, this solution makes it possible to further guarantee the uniform illumination of the sheet to further increase the precision in the detection of the defect, also unlike the devices of the state of the art, generally incapable of providing optimal illumination in the vicinity of the fabric where any defect is sought.

Preferably, in the equipment according to the invention, the processing system is further adapted to receive a plurality of reference images of the sheet from the image capture system, to extrapolate the defect defining characteristics for the tissues , classifying the defect-defining characteristics according to successive learning layers and storing these defect-defining characteristics in dedicated parts of the database.

Advantageously, the present solution makes it possible to establish an appropriate database on the characteristics which characterize an anomaly as a defect. The extrapolation of these characteristics therefore makes it possible to identify in real time a defect on a fabric before it reaches the spreading surface, a defect which can also be different from all the images analyzed, only the particularities being evaluated.

This learning step is usually done only once to train the system to recognize a fault.

Furthermore, advantageously, in accordance with the requirements described above for large-scale production, the present solution presents a high level of automation, reducing the "operator time" necessary directly on the machine, but keeping a specialized operator in charge. , even remotely, of a high level control task only.

Preferably, the processing system is also suitable for processing and storing a cut line containing a defect.

Advantageously, cutting equipment often provided downstream of the spreading equipment can follow this cutting line to eliminate the parts containing the defects and improve the correct production.

Preferably, the processing system is further adapted to map a plurality of images to form an overall image of an extended portion of sheet or the entire sheet of fabric, identifying the classified images as flaws on the overall picture.

Advantageously, this solution makes it possible to obtain an overall image of an extended part possibly containing defects, thus making it possible to evaluate the material to be cut as a whole.

More preferably, the control signal activates an audible and/or visual output upon identification of a fault by the processing system.

Advantageously, the type of signaling used can be adaptable and consistent with the actual requirements.

Preferably, in the equipment according to the invention, the control signal activates a spreading blocking device when an image is classified as a defect.

Advantageously, this characteristic represents an additional safety measure for particular conditions.

According to another aspect of the invention, there is proposed a method for identifying defects on fabrics for at least one sheet of fabric comprising the steps of setting up an image capture system in correspondence with a zone of detection located downstream of spreading strips and upstream of a spreading surface, installation of an image processing system connected to the image capture system and to at least one database, and storing the defect defining characteristics in dedicated portions of this database.

This learning step is performed only once to train the system to recognize a defect, while subsequent steps are performed each time a specific sheet needs to be analyzed.

At this stage, the method provides for the sliding of the sheet to be checked close to the image capture system, the capture of at least one image of at least one part of the sheet, the processing of the at least an image using the processing system.

This processing provides for the sub-steps of extrapolation of characteristics of the image, of comparison between the characteristics of the image corresponding to the characteristics of definition of defects, of classification of the image as a defect in the event of correspondence of the characteristics of the image with defect-defining features, or classifying the image as a defect-free part if the features of the image do not match the defect-defining features.

Preferably, the method further provides the steps of positioning in sequence a plurality of sheet parts comprising known defects in correspondence with the image capture system on a learning bench, receiving a corresponding plurality reference images of the plurality of sheet portions in a processing system from the image capture system, extrapolating defect defining features from the reference images, classifying the defect defining features according to successive learning layers in the processing system.

Preferably again, the method provides for generation of a control signal by means of an error signaling device only when the image is classified as a defect.

Advantageously, this method allows correct learning of the characteristics which define a defect and real-time detection of a defect during the sliding of the sheet, before it reaches the spreading surface.

This allows an optimization of the defect identification speed, a higher precision in the detection of the defect and a less loss of production time in the following stages of spreading and cutting.

Preferably, the method further comprises a step of processing and storing a cut line containing a defect.

Advantageously, this specification makes it possible to automatically facilitate any subsequent operations on the sheet.

Preferably, the method further comprises a step of mapping a plurality of images to form an overall image of an extended portion of the fabric sheet or the entire sheet, identifying the images classified as defects on the overall picture.

Advantageously, this specification makes it possible to obtain unambiguous information.

More preferably, the above method includes a step of blocking tissue smear by generating a control signal for a smear blocker when an image is classified as a defect.

Advantageously, this feature ensures greater availability of intervention in particular conditions.

Preferably, the steps of capturing images using the image capturing system and the step of processing the reference images are carried out using a deep neural network for object recognition.

Advantageously, this machine learning technology can effectively recognize even small fabric flaws and imperfections, including faulty stapling, tears, holes and stains, as it is not related to observation. operator vision.

Preferably, further, the step of comparing the features of the image with the defect-defining features and the step of classifying the image as a defect if the features of the image match the features definition of defects are carried out according to successive learning layers in the processing system.

Preferably, the method can also provide an automatic “cut-restart” step.

Advantageously, this specification makes it possible to automatically interconnect the identification of a fault with its elimination.

Other characteristics and advantages of the equipment and of the relative method according to the present invention will emerge from the description given below of a preferred embodiment given by way of non-limiting example with reference to the appended figures.

Brief description of figures

There schematically represents an overall view of a spreading equipment according to the state of the art, with a spread sheet of fabric;

There schematically represents another overall view of a spreading equipment according to the state of the art, with a spread sheet of fabric;

There schematically represents a spreading equipment according to a first embodiment according to the invention;

There schematically represents a spreading equipment according to a second embodiment according to the invention;

There shows a diagram of a fault identification method according to the present invention;

There represents an example of an identifiable defect according to the present invention;

There represents an identification of the fault of the .

In the various figures, similar elements will be identified by similar reference numerals.

Claims (18)

Spreading equipment (1, 1') for fabrics, comprising:
- at least one movable frame (2);
- at least two spreading strips (3);
- at least one handling means (4) for said spreading strips (3);
- a spreading structure (5) comprising a spreading surface (5A) for at least one sheet (6) of said fabrics and at least one crossbar (9);
characterized in that it comprises a system for identifying defects (10, 10') on fabrics comprising:
- at least one image capture system (11A, 11B) positioned on said crosspiece (9) facing said spreading strips (3) and adapted to capture at least one image of said sheet (6) in correspondence with an area sensor (A) located downstream of said spreading strips (3) and upstream of said spreading area (5A);
- at least one database (13) in which defect defining characteristics (18) are stored;
- at least one processing system (12) active during the spreading of said sheet (6), connected to said at least one image capture system (11A, 11B) and connected to said at least one database ( 13), said processing system (12) being suitable for:
- receiving at least one image (19) of said sheet (6) from said image capture system (11A, 11B);
- extrapolate the characteristics (24) of said image (19);
- comparing said features (24) of said image (19) with said defect defining features (18); And
- if said features (24) of said image (19) match said defect defining features (18), classifying said image (19) as a defect. Tissue spreading equipment (1) according to claim 1, further comprising an error signaling device (14) operatively connected to said processing system (12), said processing system (12) generating a control (20) for said error signaling device (14) upon matching of said features (24) of said image (19) with said defect defining features (18) and said classification of said image (19) as default. Tissue spreading equipment (1) according to claim 1 or 2, comprising a plurality of image pick-up devices (11A), arranged at a predetermined distance along said traverse (9) towards the detection zone (A ). Tissue laying equipment (1') according to claim 1 or 2, wherein said image capture system (11B) comprises at least one linescan camera, adapted to frame the entire detection area (A). Tissue spreading equipment (1, 1') according to any one of claims 1 to 4, further comprising an illumination device for said detection area (A). Tissue spreading equipment (1, 1') according to any one of claims 1 to 5, wherein said treatment system (12) is further adapted for:
- receiving a plurality of reference images (16) of said sheet (6) of said tissues from said image capture system (11A, 11B);
- extrapolating said defect defining characteristics (18) for said tissues and classifying said defect defining characteristics (18) according to successive training layers (25);
- storing said defect defining characteristics (18) in said dedicated parts of said database (13). Tissue laying equipment (1, 1') according to any one of claims 1 to 6, wherein said processing system (12) is further adapted to perform mapping of a plurality of images (19) to form an overall image of an extended part of said fabric sheet (6), by identifying the images (19) classified as defects on said overall image. Tissue spreading equipment (1, 1') according to any one of claims 2 to 7, wherein said control signal (20) activates an audible and/or visual output upon identification of a fault by said processing system (12). Tissue spreading equipment (1, 1') according to any one of claims 2 to 8, wherein said control signal (20) activates a spreading blocking device (21) when an image (19) is classified as a defect. Tissue laying equipment (1, 1') according to any one of claims 1 to 9, wherein said processing system (12) is further adapted to process and store a cut line containing a defect. Method for identifying faults on fabrics for at least one sheet (6) of fabric comprising the steps:
- preparation of an image capture system (11A, 11B) in correspondence with a detection zone (A) located downstream of the spreading strips (3) and upstream of a spreading surface (5A );
- preparation of an image processing system (12) connected to said image capture system (11A, 11B) and to at least one database (13);
- storing defect defining characteristics (18) in said dedicated parts of said database (13);
- sliding of said sheet (6) close to said image capture system (11A, 11B);
- Capture of at least one image (19) of at least part of said sheet (6);
- processing of said at least one image (19) by means of said processing system (12) according to the following sub-steps:
- the extrapolation of characteristics (24) of said image (19);
- comparing said features (24) of said image (19) with said defect defining features (18); And
- classifying said image (19) as a defect if said features (24) of said image (19) match said defect defining features (18); Or
- classifying said image (19) as a defect-free portion if said features (24) of said image (19) mismatch said defect defining features (18). A method of identification according to claim 11, further comprising the steps:
- sequential positioning of a plurality of sheet parts (15A, 15B, 15C) comprising known defects (17A, 17B, 17C) at the level of said image capture system (11A, 11B) on a learning bench ;
- receiving a corresponding plurality of reference images (16) of said plurality of sheet portions (15A, 15B, 15C) in a processing system (12) from said image capture system (11A, 11B );
- extrapolation of defect defining characteristics (18) from said reference images (16); And
- classifying said defect defining features (18) according to successive learning layers (25) in said processing system (12). 13. Identification method according to claim 11 or 12, further comprising a step of generating a control signal (20) by means of an error signaling device (14) only when said image (19) is classified as a defect. 14. Identification method according to any one of claims 11 to 13, further comprising a step of mapping a plurality of images (19) to form an overall image of an extended part of said sheet (6) of tissue, identifying the images (19) classified as defects on said global image. A method of identification according to any of claims 11 to 14, further comprising the step of blocking a spread of said tissue by generating a control signal for a blocking device (21) when an image (19) is classified as a defect. Identification method according to any one of claims 11 to 15, wherein said steps of capturing images by means of said image capturing system (11A, 11B) and said step of processing said reference images (16) are performed using a deep neural network for object recognition. An identification method according to claim 16, wherein said step of comparing said features (24) of said image (19) with said defect defining features (18) and said step of classifying said image (19) as a defect in case of correspondence of said characteristics (24) of said image (19) with said defect defining characteristics (18) are carried out according to successive learning layers (25) in said processing system (12). Identification method according to any one of Claims 11 to 17, further comprising an automatic “cut-resume” step.