AU2021103046A4

AU2021103046A4 - Intelligent detection device for fabric defects

Info

Publication number: AU2021103046A4
Application number: AU2021103046A
Authority: AU
Inventors: Xiaodong Chen; Yuxin Chen; Zelin Chen; Xiuzhuang Mei; Li Qiu; Liandong Wang; Xiaojun Yuan; Huan ZHAO
Original assignee: Inner Mongolia University of Technology
Current assignee: Inner Mongolia University of Technology
Priority date: 2021-06-03
Filing date: 2021-06-03
Publication date: 2021-07-29
Anticipated expiration: 2029-06-03

Abstract

The present invention discloses an intelligent detection device for fabric defects, including a box body. A distribution cavity is formed in the rear side of the box body; a controller is arranged inside the distribution cavity; an upper side face of the box body is fixedly connected with a bracket; a camera used for acquiring fabric images is mounted at a central position of the bracket; the acquired fabric image information is collected in the controller by the camera and transmitted into an external server system via a circuit or a network bridge; and a server transmits the acquired images into a detection system in real time so as to perform image preprocessing and defect detection. In the present invention, corresponding fabric images are acquired by detection equipment; light assistance is conducted with respect to fabrics; fabric feature maps can be well acquired under a lighting effect; and the acquired images are trained and detected by a deep learning network, thereby accurately and efficiently detecting the fabric defects. Drawings of Description 2 61 FIG. 1 3 16 613 FIG. 2 1

Description

Drawings of Description

2

61

FIG. 1

3

16 613

FIG. 2

Description

INTELLIGENT DETECTION DEVICE FOR FABRIC DEFECTS

Technical Field

The present invention relates to the technical field of fabric defects, and particularly relates to an intelligent detection device for fabric defects.

Background

At present, fabric defects are subjected to body inspection, light inspection, finish inspection and quality inspection by most of the enterprises in China and abroad in a manual manner, i.e., fabrics are placed under a light box or a garment illuminator by inspection personnel, and various defects on surfaces of the fabrics are detected by human eyes. The defects are divided into different types by setting up internal evaluation criteria by the enterprises. The manual detection has shortages as follows: (1) Defect detection is repetitive work and needs highly focused attention; the inspection personnel are easily tired during detection; serious missing inspection of complex texture defects exists; and an accuracy rate may be up to 60-70% only; (2) Manual detection cost is higher; a detection range is pretty limited; and technical levels of the staff are uneven, thereby directly affecting overall detection efficiency and quality. However, staff performance is measured by a detection amount in enterprises, which may drive the workers to lower detection quality in pursuit of quantity. Through field investigation of the enterprises, the missing inspection problem still exists after repeated manual detection. Even if a detection worker is highly skilled, the same missing inspection problem may still exist, which will directly influence economic and social benefits of the enterprises. When defects of the fabrics are detected, obstruction of efficiency improvement is an important aspect during transmission and placement of the fabrics. Therefore, an intelligent detection device for fabric defects is provided.

Description

Summary A purpose of the present invention is to provide an intelligent detection device for fabric defects, for solving problems proposed in the background. To achieve the above purpose, technical solutions of the present invention are provided as follows: the intelligent detection device for fabric defects includes a box body, wherein a distribution cavity is formed in the rear side of the box body; a controller is arranged inside the distribution cavity; an upper side face of the box body is fixedly connected with a bracket; a camera used for acquiring fabric images is mounted at a central position of the bracket; the acquired fabric image information is collected in the controller by the camera and transmitted into an external server system via a circuit or a network bridge; and a server transmits the acquired images into a detection system in real time so as to perform image preprocessing and defect detection. An annular lamp tube used for compensating a fabric image acquisition light source is mounted on an outer wall of the bracket; a placement box and a recycling box are respectively fixedly connected to both sides of the box body; a transport channel is formed inside the box body, the placement box and the recycling box; first chutes and second chutes are formed in side walls of the transport channel; a plurality of object stages used for fixedly clamping fabrics are transported inside the transport channel by virtue of a transport mechanism; a first sliding column and a second sliding column are fixedly connected to side walls of the object stages; the first sliding column and the second sliding column are respectively connected inside the first chutes and the second chutes in a sliding manner; and the object stages are in a state of horizontally supporting the fabrics while passing through the camera by virtue of match of the first sliding column, the second sliding column, the first chutes and the second chutes.

Description

Preferably, each of the object stages includes a main board, clamping grooves, glass, clamping frames and clamping devices; the glass is fixedly mounted on the main board; the clamping grooves are formed in the periphery of the glass and positioned on side walls of the main board; the clamping frames are hinged onto side walls of the main board; and the clamping frames are movably clamped inside the clamping grooves by virtue of the clamping devices. Preferably, the first chutes include a first feed chute, a feeding slot, a first deviation slot and a first accumulation slot; and the second chutes include a second feed chute, an angle-adjusting groove, a supporting groove, a reset slot, a second deviation slot and a second accumulation slot. Preferably, the transport mechanism includes a motor, a belt, a first driver block, a roller shaft and a second driver block; the belt is driven by the motor to rotate inside the transport channel; the first driver block is fixedly connected to a side wall of the belt and is used for driving the first sliding column to slide inside the feeding slot; the roller shaft is connected with an output shaft of the motor via the belt; and the second driver block is fixedly connected to the roller shaft and is used for driving the first sliding column to slide inside the first accumulation slot. Preferably, box doors are hinged to ports of the placement box and the recycling box; an access door is hinged onto a front side wall of the box body; and the box doors and the access door are all closed and sealed by permanent magnets. Preferably, a temperature sensor is fixedly mounted on an inner side wall of the box body. Preferably, acquisition of fabric images by the camera includes the following steps: Si: placing object stages inside the placement box in sequence, wherein the placement method is as follows: after the object stages are vertically placed, the first sliding column and the second sliding column on the object stages respectively slide into the first feed chute and the second feed chute; all the object

Description

stages slide into the placement box according to slope of the first feed chute and the second feed chute, i.e., the first sliding column slides towards the direction of the feeding slot, and the second sliding column slides towards the direction of the angle-adjusting groove, so that the first sliding column on the firstly placed object stage slides to stop at a joint of the first feed chute and the feeding slot, and the second sliding column slides to stop at a joint of the second feed chute and the angle-adjusting groove; S2: starting the motor; and then driving the first driver block to slide by the belt, so that the first driver block drives the first sliding column to slide towards the direction of the first deviation slot; then the second sliding column slides in the adjusting groove along with the first sliding column and gradually slides towards the supporting groove; and the object stages gradually change from a vertical state to a horizontal state; after the first object stage is driven by the transport mechanism to slide, the next object stage automatically slides, so that the first sliding column on the object stage slides to stop at the joint of the first feed chute and the feeding slot, and the second sliding column slides to stop at the joint of the second feed chute and the angle-adjusting groove; S3: enabling the object stage to be in a horizontal state after the second sliding column slides into the supporting groove along with the first sliding column; tiling fabrics on the object stage under the camera; then stopping the motor, thereby acquiring images of the fabrics by the camera; S4: starting the motor again after acquisition, so that the first driver block drives the first sliding column to continuously slide in the process, the second sliding column slides into the reset slot from interior of the supporting groove and the second sliding column slides inside the reset slot by virtue of a gravity action of the object stage, i.e., the object stage makes a circular motion around the first

Description

sliding column and then the object stage resets from a horizontal state to a vertical state; S5: driving the first sliding column to slide towards the direction of te first deviation slot by the first driver block after the object stage resets to the vertical state; sliding the second sliding column to the joint of the reset slot and the second deviation slot when the first sliding column slides to the joint of the feeding slot and the first deviation slot; and deviating the first sliding column from the first driver block when the first sliding column slides at the joint of the first deviation slot and the first accumulation slot along the slope of the first deviation slot; S6: driving the second driver block to push the first sliding column to slide away from the direction of the first deviation slot by virtue of rotation of the second driver block driven by the roller shaft; S7: driving the next object stage to repeat operations of Si-S6 by the first driver block after the belt drives the first driver block to rotate by one circle; and further automatically acquiring images of the fabrics. Compared with the prior art, the present invention has beneficial effects as follows: 1. In the present invention, corresponding fabric images are acquired by the detection equipment; by virtue of light assistance, fabric feature maps can be well acquired under the lighting effect; and the acquired images are trained and detected by a deep learning network, thereby accurately and efficiently detecting the fabric defects. Moreover, the plurality of object stages that are fixedly clamped with fabric samples are placed in the placement box; and the object stages can be automatically driven to move by virtue of the transport mechanism, so that the fabric samples on the object stages are placed under the camera one by one so as to sample the images. Therefore, complex steps of manually taking out the samples and placing the samples again are saved after image acquisition each time; degrees

Description

of intelligence and simplicity are increased; and efficiency of acquiring the fabric images is increased. 2. In the present invention, with respect to the fabric defects, the fabric images can be conveniently and automatically acquired; and effects of artificial subjective factors can be overcome by introducing an acquisition and recognition algorithm. Thus, with self-learning of a neural network, a recognition rate of the fabric defects is continuously increased.

Description of Drawings

Fig. 1 is a schematic diagram I of an overall structure in the present invention; Fig. 2 is a schematic diagram II of an overall structure in the present invention; Fig. 3 is a schematic diagram III of an overall structure in the present invention; Fig. 4 is a structural schematic diagram at a box body, a camera and an annular lamp tube in the present invention; Fig. 5 is a structural schematic diagram I at a main board, glass and a clamping frame in the present invention; Fig. 6 is a structural schematic diagram II at a main board, glass and a clamping frame in the present invention; Fig. 7 is a sectional view at a main board, glass and a clamping frame in the present invention; Fig. 8 is an enlarged view at A in Fig. 7 in the present invention; Fig. 9 is a structural schematic diagram at a first chute and a second chute in the present invention; Fig. 10 is a sectional view at a box body, a placement box, a recycling box, a camera and an object stage in the present invention; Fig. 11 is an enlarged view at B in Fig. 10 in the present invention;

Description

Fig. 12 is a structural schematic diagram at a belt, a first driver block, a roller shaft and a second driver block in the present invention; Fig. 13 is a schematic diagram I of a moving process of an object stage, a first sliding column and a second sliding column in the present invention; Fig. 14 is a schematic diagram II of a moving process of an object stage, a first sliding column and a second sliding column in the present invention; Fig. 15 is a schematic diagram III of a moving process of an object stage, a first sliding column and a second sliding column in the present invention; Fig. 16 is a schematic diagram IV of a moving process of an object stage, a first sliding column and a second sliding column in the present invention; Fig. 17 is a schematic diagram V of a moving process of an object stage, a first sliding column and a second sliding column in the present invention; Fig. 18 is an enlarged view at C in Fig. 17 in the present invention; Fig. 19 is a schematic diagram VI of a moving process of an object stage, a first sliding column and a second sliding column in the present invention; Fig. 20 is a structural schematic diagram at a box body, a breather pipe and an exhaust pipe in the present invention; Fig. 21 is a structural schematic diagram at a distribution cavity, a peltier module, a cold box and a hot box in the present invention; Fig. 22 is a section view at a peltier module, a cold box, a hot box and a Y-shaped conduit in the present invention; Fig. 23 is a schematic diagram I of a control circuit in the present invention; and Fig. 24 is a schematic diagram II of a control circuit in the present invention. In the figures: 1: box body; 101: distribution cavity; 2: bracket; 3: camera; 4: object stage; 401: main board; 402: clamping groove; 403: glass; 404: clamping frame; 405: clamping device; 4051: fixture block; 4052: clamp slot; 4053: spring; 4054: push handle; 5: placement box; 6: recycling box; 7: transport mechanism;

Description

701: motor; 702: belt; 703: first driver block; 704: roller shaft; 705: second driver block; 8: first chute; 801: first feed chute; 802: feeding slot; 803: first deviation slot; 804: first accumulation slot; 9: second chute; 901: second feed chute; 902: angle-adjusting groove; 903: supporting groove; 904: reset slot; 905: second deviation slot; 906: second accumulation slot; 10: annular lamp tube; 11: first sliding column; 12: second sliding column; 13: access door; 14: permanent magnet; : temperature sensor; 16: box door; 17: controller; 18: peltier module; 19: cold box; 20: hot box; 21: Y-shaped conduit; 22: unidirectional sealing plate; 23: fan; 24: breather pipe; 25: exhaust pipe.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and fully described below in combination with the drawings in the embodiments of the present invention. Apparently, the described embodiments are merely part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those ordinary skilled in the art without contributing creative labor will belong to the protection scope of the present invention. Referring to Figs. 1-24, the present invention provides a technical solution as follows: an intelligent detection device for fabric defects includes a box body 1. A distribution cavity 101 is formed in the rear side of the box body 1; the rear side of the distribution cavity 101 is sealed by a rear cover plate; holes through which wires extend out are formed in side walls of the rear cover plate; a controller 17 is arranged inside the distribution cavity 101; an upper side face of the box body 1 is fixedly connected with a bracket 2; a camera 3 used for acquiring fabric images is mounted at a central position of the bracket 2; the acquired fabric image information is collected in the controller 17 by the camera 3 and transmitted into an external server system via a circuit or a network bridge, wherein "a" in Fig. 24

Description

represents the external server; and the server transmits the acquired images into a detection system in real time so as to perform image preprocessing and defect detection. Detection and learning of fabric defects may be realized in the form of a neural network algorithm yolo3 by fabric defect detection. Algorithm procedures may be as follows: (1) Fabric images are segmented into mxm grids. When target centers of fabric defects drop in one grid, defect targets are detected by the grid; some target center points may drop at the boundary of multiple grids; and the defects may be screened by virtue of non-maximum suppression. (2) P bounding boxes and confidence coefficients thereof are predicted for each grid. The confidence indicates confidence of objects included in the bounding boxes and accuracy of targets predicted in the bounding box. The confidence is defined as follows: W. = argminW i1 t - W.'$(R' ) +yl W Pr(object)*IOU"" If no target is included, the confidence is equal to 0; and if there is a target, confidence = IOU"

(3) Each bounding box includes five predicted values, that is, x, y, w, h and confidence. (x, y) represents a center coordinate of the bounding box and is related to the grid; and the x, y, w and h are normalized in the training process; (4) Q conditional class probabilities Pr(classi object) are predicted for each grid; and regardless of the number of the bounding boxes in each grid, a group of class probability prediction is performed on each grid only. (5) The conditional class probabilities and the confidence of the bounding box are multiplied at a test stage: Pr(class Iobject)* Pr(object)* IOU' = Pr(class )*IOUgi

Thus, a confidence score of each bounding box within a specific class is obtained. Therefore, the final predicted value becomes a tensor of mxmx(P*5+Q).

Q

Description

Through the bounding box prediction method, an anchor box is determined by virtue of size clustering. Four coordinate offsets such as tx, ty, tw and th are predicted for each bounding box network. If a coordinate at a top left corner of an offset image in a certain unit of feature map is (qx, qy) and a preselected box size of the bounding box is bw and bh, i.e., an anchor size, then paired prediction coordinates of px, py, pw and ph are generated, that is, feature map hierarchy, while gx, gy, gw and gh are maps of truth values on the feature map. The px, py, pw and ph are consistent with the gx, gy, gw and gh by virtue of the prediction offsets of tx, ty, tw and th. P,- (t ) +QX P__a (t,) +Q, P, be'

Pih bbe'h Since feature information is decreased in a convolutional neural network training process, a Densenet structure is cited and connects each layer to other layers in a feedforward mode, thereby reserving all feature information. Thus, learning may be continuously improved by virtue of continuous data acquisition and artificial expert intervention, thereby realizing full automatic detection. The Densenet structure replaces an original low-resolution input layer by a dense network on the basis of a Darknet-53 architecture, thereby enhancing feature communication and promoting feature reuse and fusion. The input image size is changed from 256x256 to 512x512; and a sampling layer under 32x32 and 16x16 in the original network is modified as the Densenet structure. Certainly, a specific algorithm may be of other forms, and is specifically operated according to the server design. An annular lamp tube 10 used for compensating a fabric image acquisition light source is mounted on an outer wall of the bracket 2; brightness of the annular lamp tube 10 can be automatically adjusted so as to automatically change light

In

Description

intensity of the fabrics; a placement box 5 and a recycling box 6 are respectively fixedly connected to both sides of the box body 1; a transport channel is formed inside the box body 1, the placement box 5 and the recycling box 6; first chutes 8 and second chutes 9 are formed in side walls of the transport channel; a plurality of object stages 4 used for fixedly clamping fabrics are transported inside the transport channel by virtue of a transport mechanism 7; a first sliding column 11 and a second sliding column 12 are fixedly connected to side walls of the object stages 4; the first sliding column 11 and the second sliding column 12 are respectively connected inside the first chutes 8 and the second chutes 9 in a sliding manner; and the object stages 4 are in a state of horizontally supporting the fabrics while passing through the camera 3 by virtue of match of the first sliding column 11, the second sliding column 12, the first chutes 8 and the second chutes 9. As shown in Figs. 5-8, to rapidly fixedly clamp the fabric samples, specifically, each of the object stages 4 includes a main board 401, clamping grooves 402, glass 403, clamping frames 404 and clamping devices 405; the glass 403 is fixedly mounted on the main board 401; the glass 403 is transparent ground glass; the clamping grooves 402 are formed in the periphery of the glass 403 and positioned on side walls of the main board 401; the clamping frames 404 are hinged onto side walls of the main board 401; the clamping frames 404 are rotated to a state shown as Fig. 6; the fabric samples are placed on the glass 403; edges of the fabric samples are located inside the clamping grooves 402; and the clamping frames 404 are closed to fix the edges of the fabric samples inside the clamping grooves 402 by the clamping frames 404, so that the fabric samples are fixedly tiled on the glass 403. Moreover, the clamping frames 404 are movably clamped inside the clamping grooves 402 by virtue of the clamping devices 405. As shown in Fig. 8, each of the clamping devices 405 includes a fixture block 4051, a clamp slot 4052, a spring 4053 and a push handle 4054; the fixture block 4051 is connected inside the clamping frame 404 by the spring 4053 in a sliding manner;

Description

the push handle 4054 is fixedly connected with the fixture block 4051; the clamp slot 4052 is formed in a side wall of the main board 401; and after the clamping frame 404 is closed, the fixture block 4051 is clamped inside the clamp slot 4052, and thus the clamping frame 404 may be fixed on the main board 401 so as to fixedly clamp the fabrics by the clamping frame 404. When the fabric samples need to be taken down, the push handle 4054 is pushed; thus the fixture block 4051 slides out of the clamp slot 4052 by the push handle 4054; and then the clamping frame 404 may rotate on the main board 401, thereby taking down the fabric samples so as to rapidly replace the fabric samples. As shown in Figs. 9-12, to enable the object stages 4 to move inside the transport channel, specifically, the first chutes 8 include a first feed chute 801, a feeding slot 802, a first deviation slot 803 and a first accumulation slot 804; and the second chutes 9 include a second feed chute 901, an angle-adjusting groove 902, a supporting groove 903, a reset slot 904, a second deviation slot 905 and a second accumulation slot 906. The first feed chute 801 and the second feed chute 901 are formed inside the placement box 5. As shown in Fig. 9, the first feed chute 801 and the second feed chute 901 are parallel to each other and have a certain slope; the first deviation slot 803 and the second deviation slot 905 are parallel to each other and have a certain slope; and the first accumulation slot 804 and the second accumulation slot 906 are formed inside the recycling box 6. As shown in Fig. 9, the first accumulation slot 804 and the second accumulation slot 906 are parallel to each other. The transport mechanism 7 includes a motor 701, a belt 702, a first driver block 703, a roller shaft 704 and a second driver block 705; the belt 702 is driven by the motor 701 to rotate inside the transport channel; the motor 701 is fixedly mounted on an inner side wall of the box body 1; the belt 702 is supported by two supporting shafts; and any of the supporting shafts is connected with an output shaft of the motor 701. Therefore, when started, the motor 701 drives the belt 702 to rotate; the first driver block 703 is fixedly connected to a side

Description

wall of the belt 702 for driving the first sliding column 11 to slide inside the feeding slot 802; and the roller shaft 704 is connected with the output shaft of the motor 701 via the belt. As shown in Fig. 12, when started, the motor 701 drives the roller shaft 704 to rotate; and the second driver block 705 is fixedly connected to the roller shaft 704 for driving the first sliding column 11 to slide inside the first accumulation slot 804. As shown in Figs. 13-19, to switch angles in a process of moving the object stages 4, specifically, acquisition of fabric images by the camera 3 includes the following steps: Si: object stages 4 were placed inside the placement box 5 in sequence, wherein a placement method was as follows: after the object stages 4 were vertically placed, the first sliding column 11 and the second sliding column 12 on the object stages 4 respectively slid into the first feed chute 801 and the second feed chute 901; all the object stages 4 slid into the placement box 5 according to slope of the first feed chute 801 and the second feed chute 901, i.e., the first sliding column 11 slid towards the direction of a feeding slot 802, and the second sliding column 12 slid towards the direction of an angle-adjusting groove 902, so that the first sliding column 11 on the firstly placed object stage 4 slid to stop at a joint of the first feed chute 801 and the feeding slot 802, and the second sliding column 12 slid to stop at a joint of the second feed chute 801 and the angle-adjusting groove 902; S2: the motor 701 was started; and then the first driver block 703 was driven to slide by the belt 702, so that the first driver block 703 drove the first sliding column 11 to slide towards the direction of a first deviation slot 803; then the second sliding column 12 slid in the adjusting groove 902 along the first sliding column 11 and gradually slid towards the supporting groove 903; and the object stages 4 gradually changed from a vertical state to a horizontal state;

Description

after the first object stage 4 was driven by the transport mechanism 7 to slide, the next object stage 4 automatically slid, so that the first sliding column 11 on the object stage slid to stop at the joint of the first feed chute 801 and the feeding slot 802, and the second sliding column 12 slid to stop at the joint of the second feed chute 801 and the angle-adjusting groove 902; S3: the object stage 4 was in a horizontal state after the second sliding column 12 slid into the supporting groove 903 along with the first sliding column 11; fabrics were tiled on the object stage 4 under the camera 3; then the motor 701 was stopped, thereby acquiring images of the fabrics by the camera 3; S4: the motor 701 was started again after acquisition, so that the first driver block 705 drove the first sliding column 11 to continuously slide in the process; the second sliding column 12 slid into the reset slot 904 from interior of the supporting groove 903; the second sliding column 12 slid inside the reset slot 904 by virtue of a gravity action of the object stage 4, i.e., the object stage 4 made a circular motion around the first sliding column 11 and then the object stage 4 reset from a horizontal state to a vertical state; S5: the first sliding column 11 was driven to slide towards the direction of a first deviation slot 803 by the first driver block 703 after the object stage 4 reset to the vertical state; the second sliding column 12 slid to the joint of the reset slot 904 and a second deviation slot 905 when the first sliding column 11 slid to the joint of the feeding slot 802 and the first deviation slot 803; and the first sliding column 11 was deviated from the first driver block 703 when the first sliding column 11 slid at the joint of the first deviation slot 803 and a first accumulation slot 804 along the slope of the first deviation slot 803; S6: the second driver block 705 was driven to push the first sliding column 11 to slide away from the direction of the first deviation slot 803 by virtue of rotation of the second driver block 705 driven by the roller shaft 704;

Description

S7: the next object stage 4 was driven to repeat operations of Si-S6 by the first driver block 703 after the belt 702 drove the first driver block 703 to rotate by one circle; and images of the fabrics were further automatically acquired. To prevent external light coming into the box body 1 from affecting fabric pattern acquisition, specifically, box doors 16 are hinged to ports of the placement box 5 and the recycling box 6; an access door 13 is hinged onto a front side wall of the box body 1; and the box doors 16 and the access door 13 are all closed and sealed by permanent magnets 14. As shown in Fig. 3, the permanent magnets 14 are fixed on side walls of the box doors 16 and the access door 13; and the permanent magnets 14 are also arranged on corresponding side walls of the box body 1, the placement box 5 and the recycling box 6. When the box doors 16 and the access door 13 are closed, attraction is produced by the two permanent magnets 14, so that the box doors 16 and the access door 13 are fixed in a closed state. Moreover, rubber gaskets are attached to the inner side walls of the box doors 16 and the access door 13; and after the box doors 16 and the access door 13 are closed, gaps of the box doors 16 and the access door 13 may be blocked by the rubber gaskets, thereby avoiding the external light from coming into the box body 1. To detect temperature inside the box body 1 and enable an operator to know temperature conditions inside the box body 1, specifically, a temperature sensor 15 is fixedly mounted on the inner side wall of the box body 1. As shown in Fig. 3, the temperature sensor 15 is located inside the box body 1, so that the temperature inside the box body 1 can be detected. Moreover, the temperature sensor 15 is electrically connected with the controller 17; then a detected temperature value is transmitted into the controller 17 by the temperature sensor 15; and the temperature value is transferred to an external display device by the controller 17. Therefore, the temperature conditions inside the box body 1 can be known by the operator. Both the temperature sensor 15 and the controller 17 are existing

Description

technology products, so unnecessary details are not given. For example, the temperature sensor 15 may adopt a model of WZPT-01; and the controller 17 may adopt a development board of CHM Raspberry Pi 4. During fabric pattern acquisition, greater errors of the acquired images are produced due to greater temperature changes, so that fabric defect detection accuracy is lowered. Therefore, the interior of the box body 1 shall be in a temperature state. To enable a constant temperature state in the box body 1, a temperature control device is added inside the distribution cavity 101. The temperature control device includes a peltier module 18. A cold box 19 and a hot box 20 are respectively arranged at two cold and hot ends of the peltier module 18; after the peltier module 18 is electrified, gases in the cold box 19 may be refrigerated, and gases in the hot box 20 are heated. The temperature control device further includes a Y-shaped conduit 21; two ports on the lower side of the Y-shaped conduit 21 are respectively communicated with the cold box 19 and the hot box 20; fans 23 are arranged at two communication positions; when the fan 23 at the joint of the ports on the lower side of the Y-shaped conduit 21 and the cold box 19 is started, cold air inside the cold box 19 can be blown into the Y-shaped conduit 21; when the fan 23 at the joint of the ports on the lower side of the Y-shaped conduit 21 and the hot box 20 is started, hot air inside the hot box 20 can be blown into the Y-shaped conduit 21; the upper port of the Y-shaped conduit 21 is communicated with the inner cavity of the box body 1; and unidirectional sealing plates 22 are arranged on communication paths of the Y-shaped conduit 21 and the cold box 19 and the hot box 20. Assuming that a temperature range of acquired fabric images is 20°-30°, when the temperature in the box body 1 is lower than or higher than the temperature range, the temperature control device is operated, and the operating process of the temperature control device is as follows (taking an optimal temperature range °-30° of the acquired fabric images as an example in the following process):

Description

The temperature inside the box body 1 is detected by the temperature sensor in real time; and the detected temperature value is transferred into the controller 17 by the temperature sensor 15. Operating process I: when the temperature in the box body 1 received by the controller 17 is in a range of 20°-30°, the temperature control device is controlled by the controller 17 to be in a standby state, i.e., the peltier module 18 and the two fans 23 are in an off state. Operating process II: when the temperature in the box body 1 received by the controller 17 is higher than 300, the temperature control device is controlled by the controller 17 to start; and then the peltier module 18 is electrified to start, so that the temperature inside the cold box 19 is lowered by the peltier module 18. Meanwhile, the fan 23 at the joint of the cold box 19 and the Y-shaped conduit 21 is started (the other fan 23 is stopped); the cold air inside the cold box 19 is blown into the Y-shaped conduit 21 by the fan 23; and then the unidirectional sealing plates 22 close to the cold box 19 are blown open, and the unidirectional sealing plates 22 close to the hot box 20 are in a sealed state, so that the cold air inside the Y-shaped conduit 21 blows into the box body 1 for cooling. When the temperature inside the box body 1 is lowered to 250 (the temperature value can be set as any one temperature value of 20°-30°), the temperature control device is controlled by the controller 17 to be in a standby state; and when the temperature inside the box body 1 rises to be higher than 30, the temperature control device is controlled by the controller 17 to start, so as to control the temperature inside the box body 1 to be in the range of 20°-30°. Operating process III: when the temperature in the box body 1 received by the controller 17 is lower than 200, the temperature control device is controlled by the controller 17 to start; and then the peltier module 18 is electrified to start, so that the temperature inside the hot box 20 is increased by the peltier module 18. Meanwhile, the fan 23 at the joint of the hot box 20 and the Y-shaped conduit 21 is

Description

started (the other fan 23 is stopped); the cold air inside the hot box 20 is blown into the Y-shaped conduit 21 by the fan 23; and then the unidirectional sealing plates 22 close to the hot box 20 are blown open, and the unidirectional sealing plates 22 close to the cold box 19 are in a sealed state so that the hot air inside the Y-shaped conduit 21 blows into the box body 1 for heating. When the temperature inside the box body 1 is increased to 250 (the temperature value can be set as any one temperature value of 20°-30°), the temperature control device is controlled by the controller 17 to be in a standby state; and when the temperature inside the box body 1 rises to be lower than 20, the temperature control device is controlled by the controller 17 to start, so as to control the temperature inside the box body 1 to be in the range of 20°-30°. In the operating processes II and III, since the gases in the cold box 19, the hot box 20, the Y-shaped conduit 21 and the box body 1 are flowable, to ensure constant air pressures of the cold box 19, the hot box 20, the Y-shaped conduit 21 and the box body 1, breather pipes 24 communicated with outside are arranged on the cold box 19 and the hot box 20 (the breather pipes 24 achieve air change effects); an exhaust pipe 25 communicated with outside is arranged on the box body 1 and bent into a U-shaped state; and therefore, the external light can be prevented from coming into the box body 1 from the exhaust pipe 25. Operating principle: during use, the clamping frames 404 are rotated to the state as shown in Fig. 6; then the fabric samples are placed on the glass 403; the edges of the fabric samples are positioned inside the clamping grooves 402; and then the clamping frames 404 are closed, so that the edges of the fabric samples are fixed inside the clamping grooves 402 by the clamping frames 404; and then the fabric samples are fixedly tiled on the glass 403. Then, the box doors 16 on the placement box 5 are opened; the object stages 4 that are fixedly clamped with the fabric samples are arranged inside the placement box 5; and then the box doors 16 are closed so that the box body 1 is in a fully sealed state. Then, the transport

Description

mechanism 7 is started so that the object stages 4 are transported one by one by the transport mechanism 7; in the transport process, the object stages 4 are turned to support the fabric samples under the camera 3; and then images of the fabric samples are acquired by the camera 3. The acquired fabric image information is collected in the controller 17 by the camera 3 and transmitted into an external server system via a circuit or a network bridge; the server transmits the acquired images into a detection system in real time so as to perform image preprocessing and defect detection; then the detected fabric samples (the object stages 4) are transported into the recycling box 6 by the transport mechanism 7; and after the fabric samples are totally detected, the box doors 16 at the ports of the recycling box 6 are opened, so that all the object stages 4 can be taken out. Although the embodiments of the present invention have been illustrated and described, multiple changes, modifications, replacements and transformations may be made to these embodiments by those ordinary skilled in the art without departing from the principles and spirit of the present invention. The scope of the present invention is limited by claims and equivalents thereof.

Claims

1. An intelligent detection device for fabric defects, comprising a box body (1), wherein a distribution cavity (101) is formed in the rear side of the box body (1); a controller (17) is arranged inside the distribution cavity (101); an upper side face of the box body (1) is fixedly connected with a bracket (2); a camera (3) used for acquiring fabric images is mounted at a central position of the bracket (2); the acquired fabric image information is collected in the controller (17) by the camera (3) and transmitted into an external server system via a circuit or a network bridge; a server transmits the acquired images into a detection system in real time so as to perform image preprocessing and defect detection; an annular lamp tube (10) used for compensating a fabric image acquisition light source is mounted on an outer wall of the bracket (2); a placement box (5) and a recycling box (6) are respectively fixedly connected to both sides of the box body (1); a transport channel is formed inside the box body (1), the placement box (5) and the recycling box (6); first chutes (8) and second chutes (9) are formed in side walls of the transport channel; a plurality of object stages (4) used for fixedly clamping fabrics are transported inside the transport channel by virtue of a transport mechanism (7); a first sliding column (11) and a second sliding column (12) are fixedly connected to side walls of the object stages (4); the first sliding column (11) and the second sliding column (12) are respectively connected inside the first chutes (8) and the second chutes (9) in a sliding manner; and the object stages (4) are in a state of horizontally supporting the fabrics while passing through the camera (3) by virtue of match of the first sliding column (11), the second sliding column (12), the first chutes (8) and the second chutes (9).

2. The intelligent detection device for fabric defects according to claim 1, wherein each of the object stages (4) comprises a main board (401), clamping grooves (402), glass (403), clamping frames (404) and clamping devices (405); the glass (403) is fixedly mounted on the main board (401); the clamping grooves (402) are formed in the periphery of the glass (403) and positioned on side walls of the

Claims

main board (401); the clamping frames (404) are hinged onto side walls of the main board (401); and the clamping frames (404) are movably clamped inside the clamping grooves (402) by virtue of the clamping devices (405).

3. The intelligent detection device for fabric defects according to claim 1, wherein the first chutes (8) comprise a first feed chute (801), a feeding slot (802), a first deviation slot (803) and a first accumulation slot (804); and the second chutes (9) comprise a second feed chute (901), an angle-adjusting groove (902), a supporting groove (903), a reset slot (904), a second deviation slot (905) and a second accumulation slot (906).

4. The intelligent detection device for fabric defects according to claim 1, wherein the transport mechanism (7) comprises a motor (701), a belt (702), a first driver block (703), a roller shaft (704) and a second driver block (705); the belt (702) is driven by the motor (701) to rotate inside the transport channel; the first driver block (703) is fixedly connected to a side wall of the belt (702) and is used for driving the first sliding column (11) to slide inside the feeding slot (802); the roller shaft (704) is connected with an output shaft of the motor (701) via the belt (702); and the second driver block (705) is fixedly connected to the roller shaft (704) and is used for driving the first sliding column (11) to slide inside the first accumulation slot (804).

5. The intelligent detection device for fabric defects according to claim 1, wherein box doors (16) are hinged to ports of the placement box (5) and the recycling box (6); an access door (13) is hinged onto a front side wall of the box body (1); and the box doors (16) and the access door (13) are all closed and sealed by permanent magnets (14).

6. The intelligent detection device for fabric defects according to claim 1, wherein a temperature sensor (15) is fixedly mounted on an inner side wall of the box body (1).

Claims

7. The intelligent detection device for fabric defects according to claim 4, wherein acquisition of fabric images by the camera (3) comprises the following steps: SI: placing object stages inside the placement box (5) in sequence, wherein a placement method is as follows: after the object stages (4) are vertically placed, the first sliding column (11) and the second sliding column (12) on the object stages (4) respectively slide into the first feed chute (801) and the second feed chute (901); all the object stages (4) slide into the placement box (5) according to slope of the first feed chute (801) and the second feed chute (901), i.e., the first sliding column (11) slides towards the direction of the feeding slot (802), and the second sliding column (12) slides towards the direction of the angle-adjusting groove (902), so that the first sliding column (11) on the firstly placed object stage (4) slides to stop at a joint of the first feed chute (801) and the feeding slot (802), and the second sliding column (12) slides to stop at a joint of the second feed chute (801) and the angle-adjusting groove (902); S2: starting the motor (701); and then driving the first driver block (703) to slide by the belt (702), so that the first driver block (703) drives the first sliding column (11) to slide towards the direction of the first deviation slot (803); then the second sliding column (12) slides in the adjusting groove (902) along with the first sliding column (11) and gradually slides towards the supporting groove (903); and the object stages (4) gradually change from a vertical state to a horizontal state; after the first object stage (4) is driven by the transport mechanism (7) to slide, the next object stage (4) automatically slides, so that the first sliding column (11) on the object stage slides to stop at the joint of the first feed chute (801) and the feeding slot (802), and the second sliding column (12) slides to stop at the joint of the second feed chute (801) and the angle-adjusting groove (902); S3: enabling the object stage (4) to be in a horizontal state after the second sliding column (12) slides into the supporting groove (903) along with the first

Claims

sliding column (11); tiling fabrics on the object stage (4) under the camera (3); then stopping the motor (701), thereby acquiring images of the fabrics by the camera (3); S4: starting the motor (701) again after acquisition, so that the first driver block (703) drives the first sliding column (11) to continuously slide in the process, the second sliding column (12) slides into the reset slot (904) from interior of the supporting groove (903), and the second sliding column (12) slides inside the reset slot (904) by virtue of a gravity action of the object stage (4), i.e., the object stage (4) makes a circular motion around the first sliding column (11) and then the object stage (4) resets from a horizontal state to a vertical state; S5: driving the first sliding column (11) to slide towards the direction of a first deviation slot (803) by the first driver block (703) after the object stage (4) resets to the vertical state; sliding the second sliding column (12) to the joint of the reset slot (904) and the second deviation slot (905) when the first sliding column (11) slides to the joint of the feeding slot (802) and the first deviation slot (803); and deviating the first sliding column (11) from the first driver block (703) when the first sliding column (11) slides at the joint of the first deviation slot (803) and the first accumulation slot (804) along the slope of the first deviation slot (803); S6: driving the second driver block (705) to push the first sliding column (11) to slide away from the direction of the first deviation slot (803) by virtue of rotation of the second driver block (705) driven by a roller shaft (704); and S7: driving the next object stage (4) to repeat operations of S1-S6 by the first driver block (703) after the belt (702) drives the first driver block (703) to rotate by one circle; and further automatically acquiring images of the fabrics.