CN112718572A

CN112718572A - Automatic management system and method for classification of recycled paper products

Info

Publication number: CN112718572A
Application number: CN202110363011.6A
Authority: CN
Inventors: 孙俊军; 孙雪建; 石义伟; 丁吉祥; 王火红
Original assignee: Zhejiang Great Shengda Packing Co Ltd
Current assignee: Hubei Dashengda Packaging Printing Co ltd
Priority date: 2021-04-02
Filing date: 2021-04-02
Publication date: 2021-04-30
Anticipated expiration: 2041-04-02
Also published as: CN112718572B

Abstract

The application discloses a system and a method for automatically managing recycled paper product classification, which firstly carry out acquisition of paper product patterns, forms and states on the currently recycled corrugated paper products based on an image recognition mode, then when the paper product is judged to be a paper box according to the form acquisition result, the paper product is divided into an intact part and a cut part according to the state acquisition result, a paper product cutting area is determined according to the division result, then, determining a repairing condition required for repairing the carton box to meet the packaging capacity according to the paper product cutting area, then selecting a paperboard with morphological characteristics meeting the repairing condition of the current carton box from a temporary cache area near the image recognition implementing area, or selecting the carton box corresponding to the repair condition met by the paper board from the backup storage area, finally picking up the selected carton box or paper board from the temporary buffer area or the backup storage area, and conveying the carton box or paper board and the current paper product to the processing preparation area. The method increases the utilization efficiency and utilization speed of renewable resources.

Description

Automatic management system and method for classification of recycled paper products

Technical Field

The present application relates to the field of paper product classification technology, and more particularly, to an automatic management system and method for classification of recycled paper products.

Background

The paper product processing industry is one of the important industries in China, both corrugated boards for packaging boxes and paper towels serving as sanitary articles are indispensable daily articles in daily life, and most of packages of products, goods and other articles are wrapped and contained by the packaging boxes formed by paper products. After the transportation of a large number of articles is completed, the paper packaging boxes for containing the articles lose the containing function, and at this time, the waste packaging boxes need to be recycled, but in the recycling process, how to automatically classify the recycled corrugated paper, and how to perform corresponding recycling, storage and management according to the classification result are problems that need to be solved at present.

Disclosure of Invention

Based on this, in order to realize that the type of the paper products is identified automatically, and the recovered paper products are recycled fully, and the paper products of different types are stored respectively in due time, the following technical scheme is disclosed in the application.

In one aspect, the present application discloses an automatic management system for recycled paper product classification, comprising:

the form state acquisition module is used for acquiring paper pattern, form and state of the currently recycled corrugated paper based on an image recognition mode;

the cutting area dividing module is used for dividing the intact part and the cutting part of the paper product according to the state acquisition result and determining a paper product cutting area according to the dividing result when the paper product is judged to be the paper box according to the state acquisition result;

the repairing condition determining module is used for determining repairing conditions for repairing the carton to meet the requirement of packaging capacity according to the paper product cutting area;

the paper matching module is used for selecting a paper board with morphological characteristics meeting the repair conditions of the current paper box from a temporary cache area near the image recognition implementation area when the paper product is judged to be the paper box according to the morphological acquisition result, and selecting the paper box corresponding to the repair conditions met by the paper board from a backup storage area when the paper product is judged to be the paper board according to the morphological acquisition result;

the paper product picking and conveying device is used for picking up the selected paper boxes or paper boards from the temporary buffer area or the backup storage area after the paper boxes or the paper boards are selected out, and conveying the selected paper boxes or the selected paper boards and the current paper products to the processing preparation area;

and the paper product sorting module is used for sorting the current paper box or the paper board according to the paper product pattern, the form acquisition result and the paper product cutting area when the paper board meeting the repair condition cannot be selected and the corresponding paper box meeting the repair condition cannot be selected, wherein the paper product picking and conveying device is also used for conveying the paper board to the temporary cache area and conveying the paper box to the backup storage area according to the paper product sorting result.

In a possible embodiment, the form state acquiring module is a method for acquiring the form of the recycled corrugated paper based on an image recognition mode, and the method includes:

acquiring appearance images of a plurality of forward visual angles and at least one oblique visual angle of the corrugated paper product;

carrying out image segmentation on the appearance images of the plurality of forward visual angles to obtain images of all paper boards in the corrugated paper product structure;

and inputting the images of the paper boards and the images of the oblique visual angles into a paper structure recognition neural network to obtain a complete degree vector in the paper form vector.

In a possible embodiment, the form state acquiring module is a method for acquiring paper pattern and state of the recycled corrugated paper based on an image recognition mode, and the method comprises the following steps:

normalizing the images of the paper boards according to a set size, and binarizing the images obtained after normalization to obtain a plurality of binary images;

calculating the difference value between the binary image and the binary image under the corresponding view angle with each intact pattern in the image library to obtain a plurality of difference value images;

counting the number of non-zero value pixels in each difference image, and identifying the image in the image library corresponding to the difference image with the number lower than a set threshold value as a paper pattern;

and acquiring a plurality of sliding window regions of the corresponding binary images of which the number is not lower than the set threshold, respectively comparing the plurality of sliding window regions with the pattern sample regions of the intact-pattern binary image, and inputting the difference image into a classifier for defect identification to obtain a state vector when the similarity between any sliding window region and any pattern sample region is higher than the similarity threshold.

In a possible embodiment, the cutting area division module is used for dividing the intact part and the cut part of the paper product according to the obtained result, and comprises:

determining a paper product structure model corresponding to the recovered corrugated paper according to the form acquisition result;

determining a damaged paperboard with the damaged degree reaching the cutting condition according to the state acquisition result;

merging the mutually-shared damaged paperboards based on the paper product structure model to obtain a merged damaged paperboard;

and taking the boundary line between all the damaged paperboards and the adjacent non-damaged paperboards as the dividing basis of the cutting part.

In a possible implementation manner, the method for determining the paper product cutting area by the cutting area dividing module according to the dividing result includes:

respectively determining the structural damage ranges of all the damaged paperboards, wherein the structural damage range of the combined damaged paperboard is obtained by enveloping the respective structural damage ranges of the damaged paperboards which are co-edge before combination;

generating an envelope rectangle of each structural damage range, which is parallel to the corresponding boundary line;

calculating the pixel distance between each envelope rectangle and the corresponding boundary line;

and generating a paper product cutting area according to the pixel distance.

In one possible implementation, the system further comprises a term obtaining module and a term monitoring module;

the time limit acquisition module is used for acquiring the receiving time information of the currently-received corrugated paper product before dividing the intact part and the cut part when the paper product is judged to be the paper box according to the form acquisition result;

the paper picking and conveying device is also used for directly conveying the paper to the time limit storage area after the receiving time information is obtained;

the time limit monitoring module is used for monitoring the time limit of each carton in the time limit storage area, and conveying the carton of which the time difference between the receipt time and the current time exceeds a time threshold value into a preset space area to divide the intact part and the cut part.

In another aspect, the present application discloses a method for automatically managing recycled paper product classification, comprising:

acquiring paper patterns, forms and states of the currently recycled corrugated paper based on an image recognition mode;

when the paper product is judged to be a paper box according to the form obtaining result, dividing the intact part and the cut part of the paper product according to the state obtaining result, and determining a paper product cut area according to the dividing result;

determining a repairing condition required for repairing the carton to meet the packaging capacity according to the paper product cutting area;

selecting a paper board with morphological characteristics meeting the repair conditions of the current paper box from a temporary cache area near an image recognition implementation area, or selecting a paper box corresponding to the repair conditions met by the paper board from a backup storage area;

picking up the selected carton or paperboard from the temporary buffer area or the backup storage area, and conveying the carton or paperboard and the current paper products to a processing preparation area;

and when the paper boards meeting the repair condition cannot be selected and the corresponding paper boxes of which the paper boards can meet the repair condition cannot be selected, classifying the current paper boxes or the paper boards according to the paper product patterns, the form acquisition results and the paper product cutting area, wherein the paper product picking and conveying device is also used for conveying the paper boards to the temporary buffer area and conveying the paper boxes to the backup storage area according to the paper product classification results.

In one possible embodiment, the acquiring of the form of the collected corrugated paper product based on the image recognition method includes:

In a possible implementation manner, the acquiring of the paper pattern and the state of the recycled corrugated paper based on the image recognition manner includes:

In a possible embodiment, the dividing of the intact part and the cut part of the paper product according to the obtained result includes:

In a possible implementation manner, the determining the paper product cutting area according to the dividing result includes:

and generating a paper product cutting area according to the pixel distance.

In one possible embodiment, the method further comprises:

when the paper product is judged to be a paper box according to the form obtaining result, before the intact part and the cutting part are divided, the receiving time information of the currently-received corrugated paper product is obtained;

after the receiving time information is acquired, directly conveying the paper product to a time limit storage area;

and monitoring the time limit of each carton in the time limit storage area, and conveying the carton of which the time difference between the receiving time and the current time exceeds a time threshold value into a preset space area to divide the intact parts and the cut parts.

The system and the method for automatically managing the classification of the recycled paper products carry out the shape and state recognition on the recycled paper products, and roughly classifies paper products into paper boards and cartons based on the recognition result, stores the paper boards which are light in weight and convenient to access immediately, judging the repairing condition of the damaged carton, repairing the damaged carton when the carton is possibly repaired without mincing and remaking pulp, matching the cardboard with the carton, the paper boxes are utilized as well as the paper boards, the utilization efficiency and the utilization speed of the renewable resources are increased, the paper boxes which can not be matched with the paper boards temporarily are stored, and when the matched paper boards are recovered, the paper boards and the paper boxes can be repaired and reused integrally, the ground and remade paper products are reduced to the cut areas, and the production resources are saved.

Drawings

The embodiments described below with reference to the drawings are exemplary and intended to be used for explaining and illustrating the present application and should not be construed as limiting the scope of the present application.

Fig. 1 is a block diagram of an embodiment of an automatic management system for sorting recycled paper products disclosed in the present application.

Fig. 2 is a schematic illustration of corrugated paper products at various degrees of integrity.

Fig. 3 is a schematic illustration of the complete carton construction.

Fig. 4 is a schematic diagram of a binary image of a paperboard and a reference binary image.

FIG. 5 is a schematic view of the co-edge relationship of the broken paperboard.

Fig. 6 is a schematic diagram of the structural failure range.

FIG. 7 is a schematic illustration of the merging of structural damage ranges.

Fig. 8 is a schematic illustration of one of the paper cut regions generation.

Fig. 9 is a schematic diagram of another paper product cutting area generation.

Fig. 10 is a schematic view of a supplemental paperboard from the area G1 in fig. 8.

Fig. 11 is a flowchart illustrating an embodiment of a method for automatically managing recycled paper product classifications according to the disclosure.

Detailed Description

In order to make the implementation objects, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described in more detail below with reference to the drawings in the embodiments of the present application.

An embodiment of the recycled paper product sorting automatic management system disclosed in the present application is described in detail below with reference to fig. 1 to 10. As shown in fig. 1, the system disclosed in this embodiment includes a form status acquiring module, a cutting area dividing module, a repair condition determining module, a paper matching module, a paper picking and conveying device, and a paper sorting module.

The form state acquisition module is used for acquiring paper patterns, forms and states of the corrugated paper products which are currently received in an image recognition mode.

When carrying out the corrugated paper recovery, can set up a paper article recycle bin of retrieving waste paper article specially, the user places the appointed entrance with the corrugated paper, and the pattern, the form and the state discernment of corrugated paper are carried out to a space region in standing to the automatic corrugated paper with the entrance of recycle bin. The space area can be provided with a camera, images are collected through the camera, and the images are analyzed and processed, so that patterns, forms and state vectors of the corrugated paper product are obtained.

The corrugated paper recovered by the recovery station can be an independent paper board or a waste paper box, whether the paper board is a complete paper box or a paper box only containing part of the paper boards, or whether the paper box and the paper boards are intact or damaged, the corrugated paper can be recovered, but because the corrugated paper has different manufacturers and brands and different forms and states, the corresponding recovery processing means and storage positions are different, the recovered corrugated paper can be automatically classified and sorted as the raw material of the multiplexing corrugated paper according to the patterns, the forms and the states of the corrugated paper, so that the corrugated paper with different forms and states can be conveniently recovered and processed, the corrugated paper can be finally recycled, and the aim of saving production resources is fulfilled.

The pattern of the paper product refers to a pattern printed on the surface of the corrugated paper, such as a pattern of a trademark, a pattern of a brand name character, a pattern of a contained product and the like.

The form vector mainly comprises the size and integrity of the corrugated paper. Dimensions include length, width and height, which can be measured by distance measuring sensors, and for single-piece cardboard corrugated paper, height and board thickness are the same. The model of the morphology vector may be represented as: XT = { Cc, Wz }, where Cc is the size and Wz is the degree of integrity. Cc = (L, W, H), Wz may be a binary number, since the carton is the most complete state of the paper product, so the integrity of the paper product is evaluated as the constituent parts of the carton, and each bit in the binary number represents whether a paper board is present in the carton, as shown in fig. 2 and 3, assuming that the carton is composed of zb₁To b₉A total of 9 sheets, Wz is a binary number of 9 bits, and if the paper is the single sheet Si in fig. 2, Wz =1000000000, where the first bit is the bottom sheet zb₁The remaining positions are the remaining boards, i.e. Si is considered as the bottom board zb in the carton₁(ii) a If the paper product is the folded paperboard Mu in fig. 2, the folded paperboard refers to a plurality of paperboard which is not box-shaped, then Wz =111000000, i.e. three single paperboards comprising Mu are considered as the bottom paperboard zb₁And two opposite side panels zb₂And zb₃(ii) a By analogy, the cardboard-lacking carton refers to a box-shaped multi-cardboard, and the cardboard-lacking carton Ib in fig. 2 has Wz =111101110, and the 5 th and 9 th positions thereof are 0, which represents the absence of one side cardboard zb₅And a corresponding top cover zb₉(ii) a Wz =111111111 for the complete carton Cb in fig. 2.

The state vector mainly includes the damage degree of the corrugated paper, such as whether the corrugated paper has cracks, holes and the likeDamage, and the specific degree and location of damage. Thus, the model of the state vector can be represented as: ZT = { (ps)₁，zb₁)，(ps₂，zb₂)，…，(ps_n，zb_n)｝，ps_iIs the breakage degree of the ith paperboard of the corrugated paper, n is the number of the paperboards contained in the corrugated paper, ps_i= (nd or sd or cd), nd indicates no breakage, sd indicates slight breakage, cd indicates severe breakage, zb_iIs ps_iAnd at the position of the corrugated paper, wherein i is the serial number of the paper boards in the preset composition structure sequence of the corrugated paper.

The form state acquisition module obtains the form vector and the state vector in an image recognition mode, and is realized by algorithms such as edge detection, threshold segmentation, gray feature extraction, geometric feature extraction and the like. For example, the contour of a paper product in an image is recognized, the size and the integrity are obtained through the contour, a shadow part in the image is recognized and taken as a damaged area, or a depth variation area (convex or concave) in point cloud data is recognized and taken as a damaged area, and a state vector is calculated based on the damaged area.

The cutting area dividing module is used for dividing the intact parts and the cutting parts of the paper products according to the state acquisition result and determining the paper product cutting area according to the dividing result when the paper products are judged to be cartons according to the state acquisition result.

Because the carton needs to be cut and repaired, whether the paper product is a single paper board, a folding paper board, a paper lacking paper or a complete carton can be judged by the numerical value of the complete degree vector in the form vector, and only the cutting area of the paper lacking paper and the complete carton is divided. The paperboard can not be repaired into the state of the carton, and only other cartons can be repaired as parts, so the paperboard does not need to be repaired, and the cutting area division is not needed.

The intact part refers to the part which can be judged from the state vector and has good packaging capability and can be reused, and the cut part refers to the part which can be judged from the state vector and has poor packaging capability and can not be reusedPart (c) of (a). Using the complete carton structure model in FIG. 3 as an example, zb₁Being bottom cardboard, zb₂To zb₅Front, back, left and right side boards zb₆To zb₉Respectively front, back, left and right top paper boards. If the side board zb₅And top paperboard zb₉When there is a severe damage, zb₅And zb₉The remaining part is the intact part. Paper cutting area refers to the portion of the paperboard that needs to be cut, such as zb, where severe breakage may occur₅And zb₉The whole is used as a paper product cutting area.

And the repairing condition determining module is used for determining repairing conditions required for repairing the carton to meet the packaging capacity according to the paper product cutting area.

Because the paper products obtained after cutting according to the cutting area are incomplete paper products, the required paperboard specification parameters for changing the incomplete products from the incomplete unavailable state to the complete available state need to be determined, and then a proper paperboard can be found for repairing the incomplete paper products, so that the incomplete paper products are changed from the unavailable state to the complete available state.

Suppose that the current paper is cut out zb₅And zb₉For incomplete paper products, the cutting area (zb)₅And zb₉) The conditions for performing the repair are: the size of the paperboard is larger than the cutting area, and the portions of the paperboard on both sides in the horizontal direction beyond the cutting area are folded inward and fixed to zb₂And zb₃The lower side of the cardboard in the vertical direction beyond the cutting area is folded inwards and fixed to zb₁Is folded up laterally outwards in the vertical direction as a new top cover zb₉。

And the paper product matching module is used for selecting the paper boards of which the morphological characteristics meet the repair conditions of the current paper boxes from the temporary cache area near the image recognition implementation area when the paper products are judged to be the paper boxes according to the morphological acquisition results. The paper product picking and conveying device is used for picking corresponding paper boards from the temporary buffer area after the paper boxes are selected and conveying the corresponding paper boards and current paper products (current paper boxes) to the processing preparation area.

The temporary buffer area is an area for storing the paper boards, and the position of the temporary buffer area is located beside an area where the form state acquisition module acquires patterns, forms and states of the corrugated paper products, so that when the currently recycled corrugated paper is identified as the paper board, the paper board is directly placed into the buffer area for storage, and the temporary buffer area is used for matching and repairing the paper box later.

When the paper product is a paper box, after the pattern, the form and the state of the corrugated paper are obtained and the repair condition is determined, whether the temporary buffer area has the paper boards meeting the repair condition can be judged when the corrugated paper is still located at the position of the image recognition implementation area, if so, the paper product picking and conveying device immediately goes to the buffer area to pick up the corresponding paper boards, then the paper boards are placed into the paper box to be conveyed to the processing preparation area, and at the moment, the paper box and the paper boards in the paper box can be spliced after cutting to obtain a complete and available paper board.

In the determination of whether the morphological characteristics satisfy the repair condition, it is assumed that the current corrugated paper, after being cut, belongs to a cardboard box lacking a double-layer top cover plate having a size of 500mm × 600mm and a flute type AB.

In this case, a single sheet ZB1 having a size of 400mm × 500mm, a single-layer carton side sheet ZB2 having a size of 550mm × 800mm and a corrugated shape of B, and a double-layer carton side sheet ZB3 having a size of 600mm × 700mm and a corrugated shape of BC are stored in the buffer area. In this case, the ZB1 did not satisfy the form characteristic requirements because the size was not large enough and the number of layers was less than 2, the ZB2 did not satisfy the form characteristic requirements because the size was less than 2 and the flute shape was different, the ZB3 satisfied the size and the number of layers was the same as the carton, and the AB flute shape also satisfied the form characteristic requirements because the impact resistance and the cushion resistance were similar to those of the BC flute shape.

Therefore, the paper product picking and conveying device determines the position of the paper board ZB3 from the buffer area and takes the paper board ZB3 out of the buffer area, then the paper box with the paper board ZB3 is conveyed to be prepared for processing, one side of the paper board ZB3 is bent, the bending distance is 200mm, namely the bending distance exceeds the original missing size, after the paper box is cut, the bent side and the inner surface of the paper board at the side part of the paper box are fixedly connected by box pins, and the paper box is completed and the complete paper box is obtained.

The paper matching module is also used for selecting the carton corresponding to the repair condition met by the paper board from the backup storage area when the paper product is judged to be the paper board according to the form acquisition result. The paper product picking and conveying device is also used for picking up the corresponding carton from the standby storage area after the paper boards are selected out, and conveying the carton and the current paper product (current paper board) to the processing preparation area.

The process of matching the paper boards with the paper boxes is similar to the matching example, and the difference is that in the above example, the currently recovered paper products are paper boxes, so that a proper paper board needs to be searched from the cache region, and the matching of the paper boards with the paper boxes belongs to the matching of the paper boxes by the paper boards; in this example, the currently recycled paper products are paper boards, so that the paper boxes matched with the paper boards need to be searched from the storage area, and the matching of the paper boards with the paper boxes also belongs to the field of matching of the paper boxes.

The paper product classifying module is used for classifying the current paper box or the paper board according to the paper product pattern, the form obtaining result and the paper product cutting area when the paper board meeting the repairing condition cannot be selected and the corresponding paper box of which the paper board can meet the repairing condition cannot be selected. The paper product picking and conveying device is further used for conveying the paper boards to the temporary buffer area and conveying the paper boxes to the standby storage area according to the paper product classification result.

If the paper matching process cannot find a suitable matching object, the paper box needs to be conveyed to a backup storage area for storing if the currently recovered paper is a paper box, and the paper board needs to be conveyed to a temporary cache area with a short distance for storing if the currently recovered paper is a paper board.

The stored basis includes paper pattern, form acquisition result and paper cutting area, wherein the paper cutting area is only suitable for the classification of the carton and not suitable for the classification of the paperboard.

The morphology acquisition result according to which the classification is based mainly includes size parameters. The two-dimensional size of the single paperboard and the three-dimensional size of the carton lack of the single paperboard are respective classification bases, and can be briefly classified into small size, medium size, large size, column shape, flat shape, special shape and the like. The form acquisition result and the paper cutting area are also one of classification bases, different parts to be cut of different paper products are different, incomplete paper products obtained after cutting are also different, the integrity degree of the paper products after cutting is obtained according to the form acquisition result and the paper cutting area during classification, and classification storage is carried out according to the integrity degree after cutting instead of the integrity degree before cutting (namely in the current state). Because the paper products are only repaired without proper paper boards at present, after the proper paper boards exist subsequently, the paper boards and the paper boxes which are not cut at present are conveyed to a processing preparation area together in the repairing process as in the case that the matched paper boards can be found immediately for repairing, so that the paper products are classified according to the cut forms, and the paper boards are determined and matched directly according to the classified states.

Meanwhile, the paper patterns represent manufacturers or brand merchants of the paper boxes, so that the paper boxes with the same paper patterns can be stored together, and the paper boxes can be directly taken when special paper boxes of certain brand merchants need to be manufactured.

It will be appreciated that the paper cutting area may be only a portion of the cut paperboard rather than all of the cut paperboard, and that if only a portion of the paperboard is cut, the repair requirements for the paperboard will be different from the repair requirements for the entire paperboard, and therefore will need to be divided into different categories for separate processing.

The embodiment identifies the form and the state of the received paper product, roughly classifies the paper product into the paper boards and the paper boxes based on the identification result, immediately stores the paper boards which are lighter in weight and convenient to store and take, judges the repair condition of the damaged paper boxes, repairs the paper boxes when the repair is possible without mincing and remaking pulp, matches the paper boards and the paper boxes, namely utilizes the paper boxes as well as the paper boxes, increases the utilization efficiency and the utilization speed of regeneration resources, stores the paper boxes which cannot be matched with the paper boards temporarily, repairs the paper boxes when the matched paper boards are recovered, and integrally enables all the paper boards and the paper boxes to be repaired and reused, reduces the minced and remaked paper products to the cut areas, and saves the production resources.

In one embodiment, the method for acquiring the form of the recycled corrugated paper product by the form state acquisition module based on an image recognition mode comprises the following steps.

Firstly, appearance images of a plurality of forward visual angles and at least one oblique visual angle of a corrugated paper product are obtained. The appearance images acquired under the forward viewing angles mainly include basic views, for example, the forward viewing angles in the embodiment include front view, rear view, left view, right view, top view and other images with front surfaces facing the surface of the paper product, and the appearance images acquired under the oblique viewing angles mainly include oblique views, for example, axonometric views and other images with front surfaces facing the edges and corners of the paper product. It should be noted that, because the received corrugated paper products have different integrity degrees, the corrugated paper products may be single paper boards, complete cartons or other integrity degree types, and the appearance images of the paper products with different integrity degree types are also different.

And secondly, carrying out image segmentation on the appearance images of the plurality of forward visual angles to obtain images of all paper boards in the corrugated paper product structure. The image segmentation method can be implemented in various ways, for example, a threshold-based image segmentation method or an edge-based image segmentation method, and the background area is removed after segmentation, and a separate image of each paperboard of the paper product is obtained. As shown in fig. 2, for a single piece of cardboard Si, it only contains one piece of cardboard, so the image obtained after segmentation is an image that does not contain the single piece of cardboard itself; for the folded paperboard Mu, the folded paperboard Mu comprises a plurality of paperboards which are not box-shaped as a whole, so that the images obtained after the segmentation are the images of the paperboards at the corresponding visual angles; for the carton Ib with the paper plates lacking, the carton Ib comprises a plurality of paper plates which are box-shaped as a whole, so that the images obtained after the division are the images of each paper plate at the corresponding visual angle, some paper plates only appear in the image at one visual angle, and some paper plates can appear in the images at a plurality of visual angles due to the absence of other paper plates; for the complete carton Cb, images of zb2, zb3, zb4, zb5 are taken from its front, rear, left and right views, respectively, and images of zb1, zb6, zb7, zb8, zb9 are taken from a top view.

And finally, obtaining a size vector in the paper form vector based on the size of the image of each paperboard and the corresponding size proportion coefficient, and inputting the image of each paperboard and the image of the oblique visual angle into a paper structure recognition neural network to obtain a complete degree vector in the paper form vector. The calculation order of the size vector and the integrity degree vector is not sequential.

In the aspect of obtaining the size vector, the size of the paper board in the image and the actual size of the paper board have a proportional relation, the actual size of the paper board can be calculated by calculating the length and the width of a pixel area of the paper board in the image and multiplying the length and the width by a size proportional coefficient obtained through testing in advance, and then the size vector Cc in the form vector of the paper product is obtained. In addition, if the paper product is a carton, the distance between the paper product and the camera for acquiring the image is different due to the different sizes of the carton, so that the larger the volume of the carton is, the closer to the camera, the larger the pixel area in the image, due to the real volume, the larger the distance becomes, thus for the side cardboard images in the front, back, left and right views, the area of the cardboard pixel area in the four views can be calculated first, then, when multiplying by the size scaling factor, a distance compensation factor having a maximum value of 1 is also multiplied, the distance compensation factor can be determined by the horizontal direction length of the pixel region in the adjacent horizontal viewing angle view of the current viewing angle, the longer the length, the closer the carton is to the camera, the smaller the distance compensation factor, so that the product result is smaller to compensate for the increased size due to the closer distance.

For example, when calculating the actual size of the sheet zb2 in the front view, it is necessary to calculate the length of the side sheet zb5 in the right view (or left view), determine the distance compensation factor according to the length, and then take the product of the front view pixel area of the sheet zb2, the size scale factor and the distance compensation factor as the actual size of the sheet zb 2. The views of at least two adjacent horizontal perspectives are dimensioned to obtain a length L, a width W and a height H forming a carton dimension vector CC.

And in the aspect of obtaining the complete degree vector, inputting the segmented paperboard image and the oblique view into a pre-trained neural network, and outputting the classification of the paper products. As shown in fig. 2, the single cardboard Si and the folded cardboard Mu can be classified by the horizontal perspective image, and the horizontal perspective images between the single cardboard Si and the cardboard boxes are also greatly different, while the horizontal perspective images between the cardboard box Ib lacking the cardboard and the complete cardboard box Cb are very similar, at this time, the length and the direction of the inward ridge line in the oblique view are needed to be distinguished, and finally, the complete degree vector Wz of the paper product is obtained.

In one embodiment, the method for acquiring the form and the state of the recycled corrugated paper based on the image recognition by the form state acquisition module comprises the following steps.

Firstly, normalizing the images of the paper boards obtained by image segmentation of the form state acquisition module according to a set size, and carrying out binarization on the images obtained by normalization to obtain a plurality of binary images.

In this embodiment, the acquisition and classification of the paper patterns are realized by comparing with good images collected in advance and stored in the image library, so in order to facilitate the later image comparison, the size of the cardboard image needs to be normalized first, so that the cardboard image is consistent with the size of a reference image in the image library, for example, the cardboard image is adjusted to be 800 × 600 size. And then, performing binary processing on the appearance image through a preset gray threshold value, filtering the color of the corrugated paper board, if printed patterns exist on the paper board or the paper board is damaged, the color in the gray image is darker than the color of the paper board due to the darker color of the patterns and the shadow at the damaged part, and further the color is kept in the binary image. As shown in fig. 4, Q1 is a binary image of the presence of a crack (black area) on the paperboard, Q2 is a binary image of the absence of any crack on the paperboard and the printing of a pattern of the letter "TM", Q3 is a binary image of the presence of a pattern of the letter "TM" printed on the paperboard and the presence of a crack on the paperboard and the absence of interference between the cracked portion and the pattern, and Q4 is a binary image of the interference between the cracked portion and the pattern (the occurrence of a crack on the pattern). The paperboard images at other visual angles also obtain corresponding binary images according to the mode.

And secondly, calculating the difference value between the binary image and the binary image under the corresponding view angle with each intact pattern in the image library to obtain a plurality of difference images.

The image library stores images of various viewing angles of the cartons printed with patterns of various manufacturers, distributors and brand merchants in advance, and the images are used as reference standards without damage. As shown in fig. 4, wherein Q5 and Q6 are images of branded patterned cardboard of different brands. And comparing the binary images of the paper boards with all the paper board images of corresponding visual angles in the library, and calculating to obtain a plurality of different image difference values aiming at different reference images. It will be appreciated that the reference image may also be a pure white binary image for screening out cardboard that is not printed with any pattern.

Then, counting the number of non-zero value pixels in each difference image, and identifying the image in the image library corresponding to the difference image with the number lower than a set threshold value as a paper pattern. The non-zero value pixels indicate that the two images are different on the pixel point, specifically, if the collected two-value images of the paper board are completely the same as the reference two-value images of the paper board, the number of the non-zero value pixels of the difference image is 0, and if the patterns are different and/or damaged, the non-0 pixels are generated, so that the more the 0 value pixels are, the more the two images are similar, the same brand pattern on the two images is judged, and the brand of the carton in the collected image is obtained; if the number of the pixels other than 0 occupies most of the number of the pixels of the image, the similarity of the two images is low, and the brand patterns on the two images are judged to be different.

Specifically, as shown in fig. 4, the difference image between Q2 and Q5 has a small degree of dispersion, and thus the pattern printed on Q2 is determined to be the pattern of Q5.

And finally, comparing the plurality of sliding window regions of the corresponding binary images of which the number is not lower than the set threshold with the pattern sample regions of the intact-pattern binary images respectively, and inputting the difference image into a classifier for defect identification to obtain a state vector when the similarity between any sliding window region and any pattern sample region is higher than the similarity threshold.

The sliding window area is a window with a size smaller than that of the binary image, the image is slid with a set step length, for example, the 800 × 600 size binary image is slid with a window of 400 × 400 according to a step length of 100, to obtain ((800-. At this time, as shown in fig. 4, since the number of non-zero-value pixels is not lower than the set threshold value due to the excessively large broken region in Q3, and the paper pattern is not recognized, the pattern is subjected to the supplementary detection by the sliding windows in this step, and the TM word is recognized in one of the sliding windows, so that it is known that the difference image certainly contains the broken region, and the pattern is input to the classifier to perform defect recognition, and it is determined that the broken defect is an edge crack, and a state vector indicating the edge crack is obtained.

Similarly to Q4 in fig. 4, Q4 also fails to correctly recognize the TM mark because the broken region is too large, but the TM mark region is less affected by breakage, and therefore, after the sliding window region and the pattern sample region are compared, the similarity thereof is still higher than the similarity threshold, and therefore Q4 is similar to Q3, but only in this case, the broken region in the difference image of Q4 does not overlap the TM pattern, and therefore, the defect of edge cracking can be recognized by most of the shape features thereof after the input of the classifier.

In one embodiment, the method for dividing the intact part and the cut part of the paper product by the cutting area dividing module according to the obtained result comprises the following steps.

Firstly, determining a paper product structure model corresponding to the recycled corrugated paper according to the form acquisition result. And the form acquisition result comprises a complete degree vector of the corrugated paper product, and the complete degree is matched and compared with a plurality of pre-established paper product structure models to determine a current corrugated paper product matching model. For example, if the current corrugated paper product is a complete carton, the matched model should be the complete carton construction model shown in fig. 3.

And secondly, determining the damaged paperboard with the damaged degree reaching the cutting condition according to the state acquisition result. The state acquisition result comprises a damage degree vector of the corrugated paper product, so the damage degree vector is judged, if the degree is higher, the damage is serious, and the paper plate of the paper product needs to be cut. It will be appreciated that the cutting is primarily directed to folded cartons and cartons comprising multiple sheets.

And then merging the broken paperboards which are mutually shared on the basis of the paper product structure model to obtain a merged broken paperboard. Taking fig. 5 as an example, zb3 and zb7 in the figure are both broken paper boards, zb3 has a broken hole, zb7 has a broken crack, both are broken paper boards, zb3 and zb7 share one edge, and the common edge is a horizontal ridge, so that the two paper boards are combined into one broken paper board according to the combination condition, and the images of the two combined paper boards are superposed up and down according to the position relationship obtained from the form vector to obtain the image of the development figure as the combined image.

And finally, taking the boundary line between all the damaged paperboards and the adjacent non-damaged paperboards as the dividing basis of the cutting part. Continuing with the example of fig. 5, after sifting and combining, the carton of fig. 5 contains two broken sheets, one of which is formed by combining zb3 and zb7, and the other of which is zb8 sheet with a partial absence. Wherein adjacent non-broken paperboard of the combined broken paperboard is zb1, zb4 and zb5, so that the left boundary line (adjacent zb 4), the right boundary line (adjacent zb 5) and the lower boundary line (adjacent zb 1) of the combined broken paperboard are used as the dividing basis between the cut part and the non-cut part; zb8 the adjacent unbroken cardboard is zb4, so the division between the cut and uncut portions is based on the horizontal ridge between zb8 and zb 4.

On the basis of the intact part and cut part dividing method, in one embodiment, the method for determining the cut area of the paper product by the cut area dividing module according to the dividing result comprises the following steps.

Firstly, the structural damage range of all the damaged paperboards is determined respectively. The structural failure range is a range of a region in which the failure region affects the load-bearing performance, the burst strength, the edge pressure strength, and other packaging capabilities of the paper product, and the range of the region is larger than the failure region. As shown in FIG. 6, the black triangular area is the cracked part of the paperboard, and the dotted broken line in the figure is the structural failure range.

And the structural damage range of the combined damaged paperboards is obtained by enveloping the respective structural damage ranges of the common damaged paperboards before combination. Taking fig. 7 as an example, zb3 and zb7 on the left side of fig. 7 are structural failure ranges of zb3 and zb7 of the carton in fig. 5, respectively, and since they are combined, the right side of fig. 7 is the combined case, and the structural failure ranges of zb3 and zb7 are enveloped to obtain the structural failure ranges indicated by dotted broken lines.

Next, an envelope rectangle of each of the structural damage ranges, which is parallel to the corresponding boundary line, is generated. The envelope rectangle is a circumscribed rectangle of the graph, and for the circumscribed rectangle, the side line of the circumscribed rectangle needs to be parallel to the side line of the paper board where the circumscribed rectangle is located, taking fig. 6 as an example, the dashed-dotted rectangle in fig. 6 is the envelope rectangle parallel to the side line of the paper board in the structural damage range. For the combined broken cardboard shown in fig. 7, the dotted line therein is an enveloping rectangle of the combined structural breaking range.

Then, a pixel distance between each of the envelope rectangles and the corresponding boundary line is calculated. Taking fig. 6 as an example, in the paper board shown in fig. 6, the right side line of the enveloping rectangle and the right side line of the paper board are placed at a distance of 0 (coinciding with each other), and the distances of the left side line, the upper side line and the lower side line are L1, L2 and L3, respectively.

And finally, generating a paper product cutting area according to the pixel distance.

And if the pixel distance between the enveloping rectangle and one of the outermost side structure boundary lines of the carton is zero and the pixel distance between the enveloping rectangle and the side lines at two sides of the outermost side line of the carton is not lower than a set value, taking the enveloping rectangle as a paper product cutting area. As shown in fig. 8, in which the broken line region G1 is a structural damage range enveloping rectangle of the combined damaged paperboard, the outermost side line of the carton is a side line which is not co-located with other paperboard, G1 coincides with the outermost side line S1, and the side lines S2 and S3 at both sides of S1 have a certain distance not less than the set value, so that G1 can be determined as a cutting region instead of determining the whole combined damaged paperboard as a cutting region. Similarly, the broken line region G2 in fig. 8 is that the structural damage range of the top cover plate coincides with the outermost sideline S4 of the top cover plate, and the sideline S5 and the sideline S6 at both sides of S4 have a certain distance not lower than the set value, so that G2 can be determined as the cutting region for cutting. Because the distance is reserved on the two sides of the envelope rectangles of G1 and G2, a piece of paperboard can be additionally fixed at the cutting area and fixed with the two sides and the bottom respectively, and the stable structure is ensured.

And if the pixel distances between the enveloping rectangle and all the carton outermost structure boundary lines are not zero, determining the carton outermost structure boundary line corresponding to the minimum pixel distance, extending the edge perpendicular to the boundary line in the enveloping rectangle to the boundary line, and taking the extended area and the enveloping rectangle as a paper product cutting area. As shown in fig. 9, in which the broken line region G3 is a structural damage range enveloping rectangle, among four pixel distances of G3, although the pixel distance of the bottom side of the enveloping rectangle is zero, the side coinciding with the bottom side is not the outermost structure boundary line, and the outermost structure boundary line closest to the enveloping rectangle is S5 and is S1, so that S5 is selected as the outermost structure boundary line of the carton with the smallest pixel distance, and the corresponding two sides of the enveloping rectangle are extended toward the position of S5, so as to obtain a shaded region G4 in the figure, which is an extended region, and G4 and G3 together form a paper cutting region. Since the region G3+ G4 formed after the extension of G3 is similar to the case of G1 described above, the structural stability can be secured by the fixation of the supplemental cardboard to both sides and the bottom as well.

And if the pixel distance between the envelope rectangle and the boundary line of the outermost layer structures of the plurality of cartons is zero, taking the paperboard where the envelope rectangle is located as a paper product cutting area. As shown in fig. 9, in which the broken line region G5 is a structural damage range enveloping rectangle, G5 coincides with the outermost side lines S4 and S6, respectively (the pixel distance is 0), and the entire top cover plate where G5 is located is cut as a paper product cutting region. Since only one side of G5 is available and the other side does not provide structural support, it is cut out entirely and is not reinforced.

It should be noted that, when the paper product cutting area determined by the cutting area dividing module is not a whole paper board but a part of the content of the paper board, the structure of the paper board to be supplemented is also included in the cut incomplete paper product model established by the paper product modeling module, the paper product classification module also classifies the paper product model into the category of the paper board to be supplemented of the corresponding brand merchant, and determines the paper board of which the specification parameters can supplement the cutting area, so that the paper board meeting the corresponding specification parameters is selected from the cache area by the paper product picking and conveying device. As shown in fig. 10, the left side of the drawing shows a state of the carton cut from the region G1 in fig. 8, and the right side shows a state of the carton after the carton is replenished with a cardboard selected from the buffer region and having a size capable of covering the cut region and having a surplus portion capable of being fixed to the remaining portion of the carton.

In one embodiment, the system further comprises a term acquisition module and a term monitoring module. The term obtaining module is used for obtaining the receiving time information of the corrugated paper product which is currently received before the intact part and the cutting part are divided when the paper product is judged to be the paper box according to the form obtaining result. The paper product picking and conveying device is also used for directly conveying the paper product to the time limit storage area after the receiving time information is obtained.

After the form vector of the corrugated paper product which is currently recycled is obtained, the form of the paper product is judged firstly, if the paper product is a single paperboard or a folding paperboard, a carton corresponding to the repair condition met by the paperboard is selected from a backup storage area, and if the paper product is a carton lacking the paperboard or a complete carton, the carton is reserved for a certain time before the intact part and the cutting part are divided to be matched with the paperboard. The purpose of reserving lies in, if the user just receives the express delivery piece of this carton packing after, just send this carton to the recycle bin immediately and carry out paper article recovery, then because the user has the possibility of returning goods, and can carry original packing back again from the recycle bin and carry out original-pack packing and send back if returning goods for the packing of article reaches the best with article self matching degree, if temporarily assemble a carton through cutting and repairing, can't reach original-pack packing's matching degree promptly, waste material resources and time again, consequently carry out the reserving of a certain time with the carton earlier, the position of reserving is located the deadline memory area of setting up alone, the storage foundation in this memory area is arranged rather than the characteristic according to the carton body with the time foundation.

The remaining time is set from the time of receipt of the articles packaged in the carton, and after 7 days or a certain period of time, for example, if the user does not pick up the carton yet, the user will not return the articles, and at this time, the carton can be recycled, if the carton is intact, the carton can be used for packaging other articles, and if the carton is damaged or lacks a cardboard, the carton can be cut and repaired for reuse.

An embodiment of the method for automatically managing recycled paper product categories disclosed in the present application is described in detail below with reference to fig. 11, and this embodiment is a method for implementing the aforementioned embodiment of the system for automatically managing recycled paper product categories. As shown in fig. 11, the method disclosed in this embodiment mainly includes the following steps 100 to 600.

Step 100, acquiring paper patterns, forms and states of the currently recycled corrugated paper products based on an image recognition mode;

200, when the paper product is judged to be a paper box according to the form obtaining result, dividing the intact part and the cut part of the paper product according to the form obtaining result, and determining a paper product cutting area according to the dividing result;

step 300, determining a repairing condition for repairing the carton to meet the requirement of packaging capacity according to the paper product cutting area;

step 400, selecting a paperboard with morphological characteristics meeting the repair conditions of the current carton from a temporary cache area near an image recognition implementation area, or selecting a carton corresponding to the repair conditions met by the paperboard from a backup storage area;

step 500, picking up the selected carton or paperboard from the temporary buffer area or the backup storage area, and conveying the carton or paperboard and the current paper products to a processing preparation area;

and 600, when the paper boards meeting the repair condition cannot be selected and the corresponding paper boxes of which the paper boards can meet the repair condition cannot be selected, classifying the current paper boxes or the paper boards according to paper pattern, the form acquisition result and the paper product cutting area, wherein the paper product picking and conveying device is further used for conveying the paper boards to the temporary buffer area and conveying the paper boxes to the backup storage area according to the paper product classification result.

In one embodiment, the acquiring of the form of the collected corrugated paper product based on the image recognition method includes:

In one embodiment, the acquiring of the paper pattern and the state of the recycled corrugated paper based on the image recognition mode comprises:

In one embodiment, the dividing of the intact part and the cut part of the paper product according to the obtained result includes:

In one embodiment, the determining the paper product cutting area according to the dividing result includes:

and generating a paper product cutting area according to the pixel distance.

In one embodiment, the method further comprises:

The division of modules, units or components herein is merely a logical division, and other divisions may be possible in an actual implementation, for example, a plurality of modules and/or units may be combined or integrated in another system. Modules, units, or components described as separate parts may or may not be physically separate. The components displayed as cells may or may not be physical cells, and may be located in a specific place or distributed in grid cells. Therefore, some or all of the units can be selected according to actual needs to implement the scheme of the embodiment.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present application should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An automated recycled paper product sorting management system, comprising:

2. The system for automatically managing recycled paper product classifications of claim 1, wherein said form status acquisition module is configured to acquire the form of the recycled corrugated paper product based on image recognition, comprising:

3. The system for automatically managing recycled paper product classifications of claim 2, wherein said form status acquisition module is configured to perform pattern and status acquisition of recycled corrugated paper products based on image recognition, comprising:

4. The system for automatically managing recycled paper product classifications of claim 2 wherein said cut area classification module is configured to classify the paper product into a good section and a cut section based on the obtained results, comprising:

5. The system for automatically managing recycled paper product classifications of claim 4 wherein said cut area classification module determines the paper product cut area based on the classification result comprising:

and generating a paper product cutting area according to the pixel distance.

6. An automatic management method for recycled paper product classification is characterized by comprising the following steps:

7. The automatic classification management method according to claim 6, wherein the acquisition of the form of the recycled corrugated paper product based on the image recognition comprises:

8. The automatic classification management method according to claim 7, wherein the acquisition of the pattern and the state of the recycled corrugated paper products based on the image recognition comprises:

9. The automated classification management method according to claim 7, wherein the dividing of the good part and the cut part of the paper product according to the obtained result includes:

10. The automatic classification management method according to claim 9, wherein the determining of the paper product cutting area based on the division result includes:

and generating a paper product cutting area according to the pixel distance.