WO2020001481A1

WO2020001481A1 - Packaging method and system for ceramic tile by color and pattern

Info

Publication number: WO2020001481A1
Application number: PCT/CN2019/093026
Authority: WO
Inventors: 武桢
Original assignee: 广东科达洁能股份有限公司
Priority date: 2018-06-28
Filing date: 2019-06-26
Publication date: 2020-01-02
Also published as: CN109051106B; CN109051106A

Abstract

Provided is a packaging method and system for ceramic tile by color and pattern, wherein the system comprises a ceramic tile color and pattern recognition system (2), a transmitting mechanism (1), a grabbing mechanism (3) and a packaging mechanism (4); the ceramic tile color and pattern recognition system (2) comprises a camera shooting device and a detection device, and the detection device comprises a memory and a processor; the camera shooting device is electrically connected with the detection device; the processor extracts image data from the memory and divides the learning sample set into a plurality of sub-samples, and each sub-sample requires to contain a color and pattern sample, so as to form the training sample set, learn the training sample set, and detect the learning results using the test sample set, continuously optimizing the learning model.

Description

Method and system for dividing and decorating ceramic tiles

Technical field

The invention relates to the technical field of ceramic tile detection, and in particular, to a method and system for packaging and decorating ceramic tiles.

Background technique

In today's society, ceramic tiles are the most important materials for building decoration. High-grade ceramic tiles are often used in the construction of expensive and high-quality houses. However, from the beginning of the 21st century, the ceramic industry is generally under pressure from rising costs, environmental protection, and anti-dumping, especially the homogeneity of industry products, which has intensified competition among various brands in the industry. In this case, ceramic tile companies pay more and more attention to product creation and more attention to intelligent production to attract more businesses to trade.

Now more and more intelligent systems are applied in the ceramic tile industry, such as automatic paving line technology, robotic glaze spraying system, etc. Better companies use digital intelligence from raw material processing to glaze line equipment. Therefore, the ceramic tile industry's Intelligent production is gradually becoming the mainstream of production, and it is no longer a fixed production scale.

At present, the tile factory basically distinguishes the color of the tiles by the workers, and transfers them to the material frame, and then packs the tiles of the same color, so that the tiles are classified and packed. The workload is very large, the labor cost is high, and the operating environment is harsh (The tiles are hot and dusty). Existing smart tile sorting and packaging devices all require a large sample size for machine learning, and then the smart tile sorting and packaging device can classify the detected tiles according to suit color. In this process, the required sample size is even Reaching tens of thousands, this undoubtedly brings great difficulties to machine learning.

Summary of the invention

In order to solve the above-mentioned problem, one of the objectives of the present invention is to provide a method and a system for packaging tiles with different designs. By dividing tiles of different designs into a plurality of sub-samples, and each sub-sample must contain a pattern of patterns, all The sub-samples constitute a training sample set to train the tile classification packaging system, so the required sample size is greatly reduced, and the detection efficiency is improved. The specific invention content is as follows:

A method for dividing and decorating ceramic tiles, comprising the following steps:

Step 1: Select tiles of different colors as the original samples. The original samples are divided into learning samples and test samples. The set of learning samples is called the learning sample set, and the set of test samples is called the test sample set. The learning sample set includes A ₁ _{_{, A 2, A 3, A}} 4 ...... A n; the test sample comprises a set of _{_{_{B 1, B 2, B 3}}} , B 4 ...... B n;

Step 2: Collect images of the tile learning sample set;

Step 3: Divide each collected learning sample into several sub-samples. Each sub-sample needs to contain a pattern. The set of all sub-samples is called the training sample set. The training sample set includes: (a ₁₁ , a ₁₂ . ..... a _1n ), (a ₂₁ , a ₂₂ ...... a _2n ), (a ₃₁ , a ₃₂ ...... a _3n ) ... (a _n1 , a _n2 ...... a _nn );

Step 4: learning the training sample set, using the test sample set to detect the learning results, so as to continuously optimize the learning model;

Step 5: Sort, stack and pack the tested tiles according to suit color.

The learning method in step 4 can be a conventional method in the field of machine learning. According to the learning method, machine learning can be divided into supervised learning, unsupervised learning, and semi-supervised learning. According to the algorithm, it can be divided into regression algorithms, instance-based algorithms, Bayesian algorithm, clustering algorithm, artificial neural network algorithm, etc.

Preferably, the shape of each sub-sample after the segmentation in step 3 may be any shape. There is no limitation on the shape of each sub-sample, and even the shape of each sub-sample can be different, but each sub-sample must contain a color pattern. Under the premise that the size of each sub-sample is not greater than the size of the original sample, each sub-sample The size of the sample can be any size.

Preferably, the step 3 specifically includes the following two sub-steps:

Step 3.1: Segment the corners of each learning sample from the background environment to form a clear outline;

Step 3.2: Divide the learning samples with clear contours into sub-samples to form a training sample set.

First, the corners of each learning sample must be segmented from the background environment to form a clear outline to avoid interference from image impurities caused by impurities in the background environment.

The second object of the present invention is to provide a tile color matching package system, which includes a tile color recognition system, a conveying mechanism, a grasping mechanism, and a packaging mechanism; the tile color recognition system is disposed directly above the conveying mechanism, so that The number of the gripping mechanisms is several, and the gripping mechanisms are arranged on both sides of the conveying mechanism, and the gripping mechanisms are arranged downstream of the tile pattern recognition system; a package is provided near each of the gripping mechanisms. Mechanism; the grasping mechanism is electrically connected with the tile pattern recognition system; the tile pattern recognition system includes a camera device and a detection device, the detection device includes a memory and a processor; the camera device and the detection device Electrical connection; the camera device is used to collect an image of the tile original sample set, and then transfer the image data to the memory of the detection device, the processor extracts the image data from the memory and completes the following instructions:

Divide the learning sample set into several sub-samples, and each sub-sample needs to contain the design pattern to form a training sample set;

Learning the training sample set, using the test sample set to detect the learning results, and continuously optimizing the learning model;

The pattern of the detected tiles is identified, and a feedback signal is sent to the corresponding grasping mechanism.

Preferably, the imaging device is a CCD camera. CCD is the abbreviation of charge coupled device. It can change light into electric charge and store and transfer the electric charge. It can also take out the stored electric charge to change the voltage. A CCD camera can also be called a CCD image sensor.

Preferably, the suit recognition system further includes a detection switch, and the detection switch is a sensor. When the detection switch detects that a tile is present, a feedback signal from the detection switch is sent to the tile classification and packaging system.

Preferably, an auxiliary light source is further provided on the imaging device, the auxiliary light source includes at least one auxiliary light emitting element, and the auxiliary light emitting element is detachable. When light sources of different colors are required in different environments, The light emitting elements can be replaced to obtain different colors of light.

Preferably, the grasping mechanism includes a fixed frame, a swing arm, and a clamping mechanism; the swing arm is hinged to the fixed frame; and a driving swing arm is further provided between the fixed frame and the swing arm A swing driving mechanism; the clamping mechanism is disposed at the top of the swing arm.

Preferably, a rotation driving mechanism for driving the clamping mechanism to rotate is further provided at the top end of the swing arm.

Preferably, the grasping mechanism is a six-axis manipulator, and a suction cup is provided at the top of the six-axis manipulator.

The beneficial effects of the present invention:

The invention provides a tile classification and packaging method. By dividing tile learning samples of different suits into a plurality of sub-samples, and each sub-sample must contain a suit pattern, all the sub-samples constitute a training sample set for machine training, so greatly reduced. The required sample size is increased, and the detection efficiency is improved.

The invention also provides a tile color-packing packaging system, which can automatically sort and pack tiles of different colors, which greatly saves classification time, improves efficiency, saves labor and reduces labor intensity, and has a simple equipment structure and a lower enterprise. cost.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are Some embodiments of the present invention, for those skilled in the art, can obtain other drawings according to these drawings without paying creative labor.

FIG. 1 is a schematic diagram of a tile color recognition system in a tile spot color packaging system of the present invention; FIG.

FIG. 2 is a schematic structural diagram of a tile color-packing packaging system of the present invention.

detailed description

Embodiment one:

As shown in FIG. 2, a ceramic tile pattern packaging system includes a tile pattern recognition system 2, a conveying mechanism 1, a grasping mechanism 3, and a packaging mechanism 4; Above, when a tile passes through the tile pattern recognition system 2, it can be detected by the tile pattern recognition system 2. The number of the grasping mechanisms 3 is several, and the grasping mechanism 3 is used for Grab tiles of different designs, and then stack and pack them, so the number of gripping mechanisms 3 is greater than or equal to the type of tile designs; the gripping mechanisms 3 are provided on both sides of the conveying mechanism 1 to facilitate gripping the And the gripping mechanism 3 is disposed downstream of the tile pattern recognition system 2. The so-called downstream is defined by the direction of the tile conveyance, and the tiles pass through the tile pattern recognition system 2 before passing The grasping mechanism 3.

A packaging mechanism 4 is provided near each of the grasping mechanisms 3 for packaging the tiles. Specifically, the packaging mechanism 4 further includes a method for overlapping the four peripheral corner positions of the tiles with the folding edges of the paper. Aligned patting brick mechanism; lower corner mechanism for inserting packaging corner codes into the four peripheral corner positions of the tile; and hemming mechanism for folding the packaging carton along the fold line, the paving brick mechanism is also connected to a paving brick mechanism Paper feeding mechanism for automatic input of packing carton.

The grasping mechanism 3 is electrically connected to the tile pattern recognition system 2, and the tile pattern recognition system 2 feeds back a signal to the grasping mechanism 3. When the tile pattern recognition system 2 recognizes that the detected tile belongs to In a certain suit, the tile suit recognition system 2 feeds back a signal to the corresponding gripping mechanism 3. When the tile is transferred to the vicinity of the corresponding gripping mechanism 3, the gripping mechanism 3 acts to grab the tile and place it Packing is performed on the packaging mechanism 4 near the grasping mechanism 3.

As shown in FIG. 1, the tile color recognition system 2 includes a camera device and a detection device. The detection device includes a memory and a processor. The camera device is electrically connected to the detection device. The camera device is used to collect tiles. The image of the original sample set, and then transmitting the image data to the memory of the detection device, the processor extracts the image data from the memory and completes the following instructions:

The learning sample set is divided into several sub-samples, and each sub-sample needs to contain a defect pattern to form a training sample set;

The pattern of the detected tiles is identified, and a feedback signal is sent to the corresponding grasping mechanism 3.

The imaging device is a CCD camera, and CCD is an abbreviation of charge coupled device. It can change light into electric charges and store and transfer the electric charges. The stored electric charges can also be taken out to change the voltage. A CCD camera can also be called a CCD image sensor.

The tile color recognition system 2 further includes a detection switch, which is used to detect whether a detected tile appears, and if there is a tile within the detection switch detection range, the detection system is activated. If no tile appears in the detection range, the detection system is not activated, and the detection switch is a sensor.

The camera device is further provided with an auxiliary light source, the auxiliary light source includes at least one auxiliary light emitting element, the auxiliary light source is used for generating light to facilitate a better sensing image, and the auxiliary light emitting element is detachable. When light sources of different colors are required in different environments, the light emitting elements can be replaced to obtain different colors of light.

The grasping mechanism 3 includes a fixed frame, a swing arm, and a clamping mechanism; the swing arm is hinged to the fixed frame; and a swing driving the swing arm is further provided between the fixed frame and the swing arm. A driving mechanism; the clamping mechanism is provided at a top end of the swing arm. The swing driving mechanism drives the swing arm to move, so as to drive the clamping mechanism to clamp the corresponding tile, and then place it on the packaging mechanism 4 near the grasping mechanism 3. The fixing frame is used to fix the grasping mechanism. Take mechanism 3, the clamping mechanism is used for clamping tiles.

A rotation driving mechanism for driving the clamping mechanism to rotate is further provided at the top end of the swing arm. The rotation driving mechanism can drive the clamping mechanism to rotate, so that the rear surface and the front surface of the tile can be replaced during the process of gripping the tiles, and the tiles can be placed in a uniform direction.

The grasping mechanism 3 is a six-axis manipulator, and a suction cup is provided at the top of the six-axis manipulator. Use the suction cup to suck up the tiles.

The invention also provides a tile defect detection method, which includes the following steps:

Step 2: Collect images of the tile learning sample set;

Step 5: Sort, stack and pack the tested tiles according to suit color.

For example, there are currently 6 tiles of different colors as the original sample set. Each tile with the color is called the original sample. The original sample is divided into learning samples and test samples. The set of learning samples is called the learning sample set. The set is called the test sample set, so the number of learning sample sets is 3 and the number of test sample sets is 3. If the learning sample is divided into 4 sub-samples, each sub-sample is rotated 90 °, 180 °, 270 ° respectively. Then, new sub-samples are formed, and the new sub-samples form a training sample set, and the number of training sample sets is 3 * 4 * 4 = 48. This will greatly reduce the sample size of the original sample.

The size of each sub-sample after segmentation in step 3 is not greater than the size of the original sample. The shape of each sub-sample can be any shape. There is no limitation on the shape of each sub-sample, and even the shape of each sub-sample can be different, but each sub-sample must contain a color pattern. Under the premise that the size of each sub-sample is not greater than the size of the original sample, each sub-sample The size of the sample can be any size.

The step 3 specifically includes the following two sub-steps:

The segmented training sample set is machine-learned in the processor of the processing device. In order to facilitate machine learning to get as many patterns as possible, the segmented sub-samples must be as small as possible, but each sub-sample must contain a pattern. It cannot be blank, otherwise it is an invalid sub-sample. After the machine learns as many patterns as possible, it uses the test sample set to detect the learning results, so as to continuously iteratively optimize the learning model. Finally, the detected patterns captured by the camera device Compared with the learned pattern, the built-in algorithm of the processor is used to determine which pattern is the detected tile.

In the embodiment of the present invention, the method for cutting a sub-sample is to set a fixed-size sliding window and slide from left to right and from top to bottom with a certain step size to segment the collected tile image, the step length The distance to move with a square, rectangle, or circle as the center. The step size is determined according to the information of the image. When the range of the defect is large, the step size should be shorter to segment more sub-samples. When the range of the defect is relatively small, the step size can be longer.

Based on the disclosure and teachings of the foregoing specification, those skilled in the art to which the present invention pertains may also make changes and modifications to the above embodiments. Therefore, the present invention is not limited to the specific embodiments disclosed and described above, and some modifications and changes to the invention should also fall within the protection scope of the claims of the present invention. In addition, although some specific terms are used in this specification, these terms are just for convenience of explanation and do not constitute any limitation to the invention.

Claims

A method for dividing and decorating ceramic tiles, comprising the following steps:

Step 1: Select tiles of different colors as the original samples. The original samples are divided into learning samples and test samples. The set of learning samples is called the learning sample set, and the set of test samples is called the test sample set. The learning sample set includes A 1 , A 2, A 3, A 4 ...... A n; the test sample comprises a set of B 1, B 2, B 3 , B 4 ...... B n;

Step 2: Collect images of the tile learning sample set;

Step 3: Divide the collected learning samples in each learning sample set into several sub-samples. Each sub-sample needs to contain a pattern. The set of all sub-samples is called the training sample set. The training sample set includes: (a 11 , A 12 ...... a 1n ), (a 21 , a 22 ...... a 2n ), (a 31 , a 32 ...... a 3n ) ...... (a n1 , a n2 ...... a nn );

Step 4: learning the training sample set, using the test sample set to detect the learning results, and iteratively optimizing the learning model;

Step 5: Sort, stack and pack the tested tiles according to suit color.
The method according to claim 1, wherein the shape of each of the sub-samples divided in step 3 can be any shape.
The method according to claim 1, wherein the step 3 specifically comprises the following two sub-steps:

Step 3.1: Segment the corners of each learning sample from the background environment to form a clear outline;

Step 3.2: Divide the learning samples with clear contours into sub-samples to form a training sample set.
A tile sub-packaging packaging system using the tile sub-packing packaging method according to any one of claims 1 to 3, comprising a tile pattern recognition system, a conveying mechanism, a grasping mechanism, and a packaging mechanism; the tile pattern The identification system is disposed directly above the conveying mechanism, the number of the grasping mechanisms is several, and is disposed on both sides of the conveying mechanism, and the grasping mechanism is disposed downstream of the tile pattern recognition system; A packaging mechanism is provided near each of the grasping mechanisms; the grasping mechanism is electrically connected to the tile pattern recognition system; the tile pattern recognition system includes a camera device and a detection device, and the detection device includes a memory And a processor; the camera device is electrically connected to the detection device; the camera device is used to collect an image of a tile original sample set, and then transfer the image data to the memory of the detection device, and the processor retrieves Extract image data from the memory and complete the following instructions:

Divide the learning sample set into several sub-samples, and each sub-sample needs to contain the design pattern to form a training sample set;

Learning the training sample set, using the test sample set to detect the learning results, and continuously optimizing the learning model;

The pattern of the detected tiles is identified, and a feedback signal is sent to the corresponding grasping mechanism.
The system of claim 4, wherein the camera is a CCD camera.
The tile color matching packaging system according to claim 4, wherein the tile color recognition system further comprises a detection switch, and the detection switch is a sensor.
The system of claim 4, wherein the camera device is further provided with an auxiliary light source.
The ceramic tile color-packing packaging system according to claim 4, wherein the grasping mechanism comprises a fixed frame, a swing arm, and a clamping mechanism; the swing arm is hinged to the fixed frame; and the fixing machine A swing driving mechanism for driving the swing arm is further provided between the frame and the swing arm; the clamping mechanism is provided at the top of the swing arm.
The system of claim 8, wherein a rotation driving mechanism for driving the clamping mechanism to rotate is further provided on the top end of the swing arm.
The system of claim 4, wherein the grasping mechanism is a six-axis manipulator, and a suction cup is provided at the top of the six-axis manipulator.