JP7328011B2 - Electronic/electrical device parts scrap composition analysis method, electronic/electrical device parts scrap composition analysis device, and electronic/electrical equipment parts scrap processing method - Google Patents

Description

TECHNICAL FIELD The present invention relates to an image extraction processing method of an object, a composition analysis method of electronic/electrical equipment parts scrap, an electronic/electrical equipment parts scrap composition analysis apparatus, and an electronic/electrical equipment parts scrap processing method.

Due to the spread of IOT devices and the evolution of AI, it has become possible to efficiently and effectively collect, share, analyze, and utilize a large amount of diverse data. In particular, the technology for processing and utilizing information obtained from images has improved dramatically.

Information processing obtained from images is used in various industries, and in recent years, from the viewpoint of resource conservation, shredder dust obtained by crushing automobiles, waste home appliances, electronic and electrical equipment parts scraps such as PCs and mobile phones It is also attracting attention in the metal recycling industry that recovers valuable metals from waste.

As a metal recycling method for recovering valuable metals from electronic and electrical equipment parts scraps, for example, after incinerating electrical and electronic parts scraps, they are pulverized into a predetermined size, and the pulverized electrical and electronic parts scraps are placed in a copper smelting furnace. A method of recovering copper by treatment is known (see, for example, Patent Document 1).

JP 2015-123418 A

However, due to the increase in the processing amount of electronic and electrical equipment parts scraps, some of the substances contained in the electronic and electrical equipment parts scraps are unfavorable substances (smelting inhibitors) for the subsequent copper smelting process. You may be charged more than before. As a result, there arises a situation in which the amount of input of electronic and electrical equipment scraps must be restricted.

From the viewpoint of efficiency, it is desirable to reduce the smelting inhibitors from the electronic/electrical equipment parts scrap in order to limit the input amount of the electronic/electrical equipment parts scrap. However, it is difficult to mechanically remove the target substances by using a sorter or the like because the electronic/electrical device parts scrap contains parts of various shapes and types. The raw material composition often changes due to differences in suppliers. Therefore, it is considered desirable from the viewpoint of efficiency if the objects to be sorted can be extracted using an image analysis technique before sorting.

However, the image analysis technique may cause misrecognition more or less. For example, crushed objects such as mounted parts, substrates, and housing plastics that constitute electronic and electrical equipment parts scraps are mixed in different shapes and areas, and therefore image recognition may be difficult in some cases. The occurrence of erroneous recognition is a problem not only in the processing of electronic and electrical equipment parts scraps described above, but also in the image recognition processing of images in which objects having various shapes and areas coexist.

In view of the above problems, the present disclosure provides an image extraction processing method of an object, a composition analysis method of electronic / electrical equipment parts scrap, an electronic / electrical equipment parts scrap composition analysis apparatus, and a Provided is a method for processing electronic and electrical equipment parts scrap.

In one aspect of the embodiment of the present invention, the recognition frame assigning means selects a plurality of objects having different shapes or areas from captured images, based on classification data including object recognition information, extracting an image of an object by adding a recognition frame; extracting an object with two or more overlapping recognition frames from the image; If the ratio of the area of the overlapping portion to the area of the recognition frame with the smaller area out of any two recognition frames is equal to or greater than the reference value, the recognition of the smaller one A method of extracting and processing an image of an object including removing a frame.

In another aspect of the embodiment of the present invention, a recognition frame of a drawing condition that is different for each of a plurality of component scraps from an image obtained by picking up electronic/electric device component scraps including a plurality of component scraps having different shapes or areas. , and extracts from the image parts scraps with two or more overlapping recognition frames, calculates the area of the overlapping portion of any two recognition frames included in the extracted parts scraps, and calculates the area of the overlapping portion of any two recognition frames When the ratio of the area of the overlapping portion to the area of the smaller recognition frame among the recognition frames is greater than or equal to the reference value, the smaller recognition frame is deleted, and the recognition frame included in the deleted image is attached. The plurality of scrap parts are classified into each of the plurality of scrap parts, the total area of each is measured, and the assumed weight per unit area of the plurality of scrap parts is multiplied by the total area of the plurality of scrap parts to obtain the number of parts. A method for analyzing the composition of electronic/electrical equipment parts scrap, including estimating the composition of the electronic/electrical equipment parts scrap by analyzing the weight ratio of the scrap.

In still another aspect of the present invention, from among images obtained by imaging electronic/electrical device parts scraps including a plurality of parts scraps having different shapes or areas, different drawing conditions are set for each of the plurality of parts scraps. A storage device having classification data including recognition information of scrap parts for giving recognition frames, a recognition frame giving means for giving recognition frames to a plurality of scrap parts based on the classification data, and two or more recognition frames. Overlapping scrap parts are extracted from the image, the area of the overlapping portion of any two recognition frames included in the extracted scrap parts is calculated, and the smaller area of the arbitrary two recognition frames is recognized. Correction means for correcting a recognition frame by deleting the smaller recognition frame when the ratio of the area of the overlapping portion to the area of the frame is equal to or greater than a reference value, and the recognition frame included in the image after deletion. A measuring means for classifying the attached plurality of scrap parts into each of the plurality of scrap parts and measuring the total area of each, and multiplying the estimated weight per unit area of the plurality of scrap parts by the total area of the plurality of scrap parts. and analyzing means for estimating the composition of the electronic/electrical equipment scrap by analyzing the weight ratio of a plurality of electronic/electrical equipment scraps.

In still another aspect of the embodiment of the present invention, an imaging step of imaging electronic/electrical device parts waste composed of a plurality of component wastes, and a plurality of component wastes from the image obtained by imaging the electronic/electrical device parts waste A recognition frame for extracting images of a plurality of component scraps and giving a recognition frame to each of the plurality of component scraps based on the classification data for storing classification data for extracting the images of the parts and classifying them for each of the plurality of component scraps In the addition step, component scraps having two or more overlapping recognition frames are extracted from the image, the area of the overlapping portion of any two recognition frames included in the extracted component scraps is calculated, and any two arbitrary recognition frames are extracted. a correction step of correcting the recognition frame by deleting the smaller recognition frame when the ratio of the area of the overlapping portion to the area of the smaller recognition frame among the recognition frames is greater than or equal to a reference value; A measurement process in which the multiple scrap parts with recognition frames attached to the image after the correction process are classified by multiple scrap parts and the total area of each is measured, and an assumption per unit area of the multiple scrap parts Processing of electronic/electrical device parts scrap, including an analysis step of analyzing the composition of electronic/electrical device parts scrap by multiplying the weight by the total area of the plurality of parts scraps and analyzing the weight ratio of the plurality of parts scraps. The method.

According to the present disclosure, an image extraction processing method for an object, a composition analysis method for electronic/electrical equipment parts scrap, an electronic/electrical equipment parts scrap composition analysis apparatus, and an electronic/electrical equipment scrap that can improve recognition accuracy It is possible to provide a method for processing equipment component waste.

BRIEF DESCRIPTION OF THE DRAWINGS It is the schematic which shows the processing apparatus of the electronic and electric equipment component waste which concerns on embodiment of this invention. FIG. 2(a) shows an example in which two recognition frames are attached to a single object, and FIG. 2(b) shows an example in which three or more recognition frames are attached to a single object. It is an explanatory view showing . It is the photograph which shows the example of the image which pinpointed the electronic/electric equipment component waste which exists in an image for every component waste, and classified each component waste with the recognition frame, respectively. It is a table|surface showing an example of the analysis result using the processing apparatus of the electronic/electrical equipment components waste which concerns on embodiment of this invention. It is a flowchart showing an example of the image extraction processing method of the target object which concerns on embodiment of this invention. BRIEF DESCRIPTION OF THE DRAWINGS It is a flowchart showing an example of the processing method of the electronic/electrical equipment component waste using the electronic/electrical equipment component waste processing apparatus which concerns on embodiment of this invention.

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. It should be noted that the embodiments shown below are examples of devices and methods for embodying the technical idea of the present invention. It does not specify anything.
In the embodiment shown below, the case of extracting electronic/electrical equipment parts scrap as an image extraction object is shown as an example. Of course, it can be applied as

(Object image extraction processing method)
An object image extraction processing method according to an embodiment of the present invention uses an analysis device to extract classification data containing object recognition information from images obtained by capturing a plurality of objects having different shapes or areas. extracting an image of a target object by giving a recognition frame to the target object, extracting a target object having two or more overlapping recognition frames from the image, and extracting an arbitrary object included in the extracted target object. The area of the overlapping portion of the two recognition frames is calculated, and if the ratio of the area of the overlapping portion to the area of the recognition frame with the smaller area out of any two recognition frames is equal to or greater than the reference value, the smaller one is selected. including deleting the recognition frame of

The object according to this embodiment is preferably a crushed object. Shredded material refers to articles that have undergone a crushing process, such as in a crusher, and that are discrete in various shapes and surface areas. In particular, when an article before crushing is composed of a plurality of parts and materials, it has the characteristic that it is difficult to completely separate each part and each material by crushing.

For example, when the object is electronic and electrical equipment parts scrap, it is typically preferable to separate substrates, plastics, metal pieces, copper wire scraps, capacitors, IC chips, and other parts scraps as individual parts. . However, in the separation by crushing, it is difficult to completely separate individual components, and there are cases where other component waste remains on top of some component waste. For example, capacitors and IC chips retain their original shape even after being separated by crushing, and many of them have the same shape and size. They also vary in size, and few have the same shape and size.

In this embodiment, the image of the object is extracted by adding a recognition frame to the object based on the classification data including the recognition information of the object. For example, when electronic / electrical equipment parts scrap is the object, using the image of electronic / electrical equipment parts scrap, for example, when classifying into three types of board, plastic, other parts scrap, board and Basic information (hereinafter also referred to as "teaching data") including information on constituent elements and colors of objects to be classified for classifying and extracting plastics and other parts waste is given, and extraction processing is performed based on this basic information. .

Here, the "components" included in the basic information include the elements that constitute each piece of scrap and the positional relationship between the elements. For example, when electronic and electrical equipment parts scraps are targeted, the constituent elements for the board include the resin layer that covers the surface of the board, the wiring layer formed on the surface of the board, and the parts that are mounted on the board. , and its size and positional relationship. The “color” included in the basic information indicates the color of the component scrap, and for example, in the case of a substrate, includes the color of a layer applied to the surface of the substrate, such as green.

For example, in the case of electronic/electrical device parts scrap as objects, such training data is prepared by manually sorting the electronic/electrical device parts scraps in advance, and sorting them into substrates, plastics, metal pieces, etc. The image is read into the analysis device and the storage device included in the analysis device. The shape and area information of the board, plastic, and metal piece read as training data are representative examples, and the shape and area are different, and there are no identical pieces. The analysis accuracy increases as the number of training data increases.

When extracting an image of an object, the features of the object in the image are compared with the respective features derived from the teacher data to be classified, and the degree of certainty, that is, the target for the features derived from the teacher data. A degree indicating how well the object matches the feature is calculated. A different recognition frame is given to each classification target, with those having certainty values or more as targets for classification.

However, in the case of a plurality of objects having different shapes or areas, a plurality of recognition frames may be set for one object. For example, if a metal piece remains on the substrate, the recognition frame for the substrate and the metal frame for the metal piece may be set in a state of being overlapped. In this case, the degree of certainty as a substrate is high, and the degree of certainty as a metal piece is also high, and the setting of the recognition frame is correct.

In addition to setting one recognition frame for the entire board, there are cases where a recognition frame as another board is added to a part of the board where there is a match as a board derived from teacher data. be. As a result, a plurality of recognition frames may be attached to one piece of scrap. Furthermore, since the substrate has a part where no circuit is arranged, the part may be marked as plastic with a recognition frame. As described above, even if the accuracy of the training data is improved and the recognition frames are appropriately assigned, the actual number of scrap parts and the number of recognition frames may deviate greatly.

Therefore, when recognizing images of substances having at least different shapes or areas, such as scraps of electronic and electrical equipment parts, as objects according to the present embodiment, the accuracy of classification data including object recognition information can be evaluated by machine learning. This is particularly advantageous in applying the processing method according to the present embodiment because misrecognition is more likely to occur than in the case of image recognition of objects having relatively the same shape and area, even if the accuracy is further enhanced. .

Regarding the degree of confidence, if the degree of certainty is set low, the number of objects for which multiple recognition frames are set increases, whereas if the degree of certainty is set high, the number of objects for which no recognition frame is set increases. There is fear. It is preferable for a person skilled in the art to appropriately change the setting of the degree of confidence according to the properties of the object.

As will be described in detail below, according to the image extraction processing method of the object according to the embodiment of the present invention, when a plurality of recognition frames are set, the overlap of the recognition frames is large. When the area of the overlapped portion of the recognition frames is larger than the area of the smaller recognition frame out of the two recognition frames, the smaller recognition frame is deleted to reduce the actual number of recognition frames. The number of objects can be made closer, and the recognition accuracy can be improved.

In this embodiment, the target object will be described by taking electronic and electrical equipment parts scraps as an example, but it is not limited to this, and crushed objects such as assemblies composed of many parts and materials Of course, it can also be applied to

(Processing device)
As shown in FIG. 1, the processing apparatus for electronic/electrical equipment parts scrap according to the embodiment of the present invention includes an imaging device 12 for imaging electronic/electrical equipment parts scrap, and an electronic/electrical equipment parts scrap estimating composition. and a sorter 13 for sorting out specific scrap parts from electronic/electrical equipment scrap parts based on the composition analysis result analyzed by the composition analysis apparatus 10 .

"Electronic/electrical device parts scrap" in the present embodiment refers to scraps obtained by crushing electronic/electrical devices such as waste home appliances, PCs, mobile phones, etc. After being collected, they are crushed to an appropriate size. point to something In the present embodiment, the crushing to obtain electronic/electrical device parts scraps may be performed by the processor himself/herself, or the crushed parts purchased in the market may be used.

The crushing method is not limited to a specific device, and may be a shearing method or an impact method. Therefore, it does not include equipment belonging to the category of grinders whose purpose is to grind finely.

Electronic and electrical equipment parts scrap consists of multiple parts scrap such as plastics (synthetic resins) used for substrates, housings, etc., metal pieces, copper wire scraps, capacitors, IC chips, and others. can be classified in more detail. Although it is not limited to the following, in the present embodiment, it is possible to suitably process electronic and electrical equipment component scraps that have been crushed to a particle size of 50 mm or less.

The composition analysis apparatus 10 can include a processing device 100 for processing composition analysis processing, a storage device 110 for storing information necessary for various controls, an input device 120 and a display device 130 . The processing device 100 includes image extraction means 140, classification means 101, measurement means 102, analysis means 103, operating condition generation means 104, change information generation means 105, position information output means 106, machine learning means 107, and update means 108. be able to.

The storage device 110 can include classification data storage means 111 , analysis information storage means 112 , operating condition storage means 113 and position information storage means 114 . The analysis means 103 can output the analysis result of the analysis means 103 through the network 11 to the server 15 or the sorting machine 14 connected through the network 11 , which is different from the sorting machine 13 .

The classification data storage means 111 stores classification data for extracting images of a plurality of component scraps, which are objects of image analysis, from among the captured images of the electronic/electrical equipment component scraps and classifying them into a plurality of component scraps. remembered. For example, the classification data storage means 111 can classify a plurality of component scraps, that is, substrates, plastics, metal pieces, copper wire scraps, capacitors, IC chips, and others (connectors, film-like components basic information (teaching data) including recognition information including constituent elements and colors of classification objects for classifying objects into at least three or more, preferably seven or more types.

As the board information, taking the board as an example, information such as the color of the resin board constituting the board, the shape and color of ICs and wiring mounted on the surface, the relative positional relationship of these, and the like is stored. In addition, the conditions for setting a specific part in the electronic/electrical device parts scrap as one of the above parts scraps can be set in advance by the operator according to the purpose of the subsequent sorting process.

The image extracting means 140 can comprise a recognition frame adding means 141 and a correcting means 142 . Recognition frame providing means 141 assigns a specific recognition frame to a target object based on classification data including target object recognition information from images obtained by imaging a plurality of target objects having different shapes or areas. Extract images of objects.

For example, the recognition frame providing means 141 can provide a minimum circumscribing figure that circumscribes the object based on the classification data including recognition information including the components and colors of the object. Various shapes such as rectangles, parallelograms, trapezoids, squares, polygons, and hyperpolyhedrons can be adopted as the minimum circumscribed figure according to the operator's request. Generally, rectangles are preferable. Used.

FIG. 2(a) shows an example of an image in which recognition frames 1 and 2 formed of minimum circumscribing rectangles are attached to the object O1 based on the classification data including the recognition information of the object O1. In FIG. 2A, two recognition frames are attached even though the number of objects O1 is one. The correcting means 142 extracts an object O1 having two or more recognition frames 1 and 2 overlapping each other from the image, and calculates the area of the overlapping portion of any two recognition frames 1 and 2 included in the extracted object O1. S _o (not shown) is calculated. Then, the ratio (area ratio S o _/ S _s ) of the area S _o (not shown) of the overlapping portion to the area S _s of the recognition frame 2 having the smaller area among the two arbitrary recognition frames 1 and 2 is the reference. If the value is equal to or greater than the value P, correction is performed to delete the smaller recognition frame 2 .

For example, in the case of FIG. 2A, the area S _o of the overlapping portion of the recognition frames 1 and 2 is the same as the area S _s of the recognition frame 2 which has the smaller area among the two recognition frames 1 and 2. Therefore, the area ratio S _o /S _s is 1.0. Here, when the reference value P is predetermined to be 0.90, the correcting means 142 deletes the recognition frame 2 because the area ratio S _o /S _s ≧reference value P holds. As a result, one appropriate recognition frame 1 is attached to the object O1.

In image recognition, as shown in FIG. 2A, a plurality of recognition frames 1 and 2 may be attached to one object O1 regardless of the accuracy of classification data. In particular, in the case of object scraps including various shapes and configurations such as electronic and electrical equipment scraps, a huge amount of correct data (teacher data) for recognizing the object O1 is held as classification data. Even in such cases, erroneous recognition occurs at a certain rate.

According to the correction means 142 according to this embodiment, an object O1 having two or more recognition frames overlapping with each other is extracted from the image attached with the recognition frame, and the number of recognition frames is determined based on the original number of the objects O1. Therefore, it is possible to reduce erroneous recognition and improve the accuracy of image recognition. can.

As shown in FIG. 2B, there are cases where three or more recognition frames 3 to 6 are added to one object O2. In this case, when three or more of the recognition frames 3 to 6 overlap, the correcting means selects two arbitrary recognition frames 3 to 6 among the large and small overlapping recognition frames 3 to 6, and based on the reference value P, It is preferable to repeat the process of deleting the smaller recognition frames 3 to 6 until there are no recognition frames 3 to 6 to be deleted. As a result, the number of recognition frames can be corrected to match the original number of objects O1.

The reference value P is expressed by the ratio (N _o /N _c ) between the actual number N _o of the objects O1 and O2 in the image and the recognition frame N _c given based on the classification data. is preferably set based on the false recognition rate (FR) of In addition to the recognition frames 1 to 6 illustrated in FIGS. 2(a) and 2(b), the recognition frames N _c assigned based on the classification data include recognition frames surrounding substances other than the target object. , a recognition frame that incorrectly encloses the background of the image, and the like. By setting the correction of the recognition frame based on the specific ratio of the occurrence of erroneous recognition of the image of the object, the correcting means 142 can perform the processing of deleting the recognition frame with higher accuracy. .

The reference value P can be appropriately set by the operator according to the object to be subjected to image analysis. The reference value P is preferably 0.75 or more and 1.0 or less, more preferably 0.85 or more and 1.0 or less, and still more preferably 0.90 or more and 1.0 or less. In particular, according to the study of the present inventor, it was found that even if the classification data were optimized, an error recognition rate of about 10 to 15% could occur in the image recognition of electronic/electrical equipment parts scraps. Therefore, when extracting a plurality of component scraps from electronic/electrical device component scraps and assigning recognition frames 1 to 6, the reference value P should be 0.85 or more, more preferably 0.90 or more. is preferred.

The setting of the reference value P can also be adjusted according to the degree of certainty. In the present embodiment, a recognition frame is assigned to a substance whose confidence exceeds a specified value. However, if the specified confidence value is set too high, the recognition frame may not be appropriately assigned, and erroneous recognition may occur. On the other hand, if the specified value of confidence is set too low, multiple recognition frames may be given to one substance, or misrecognition may occur due to overlapping recognition frame settings for one substance. . Therefore, it is necessary to appropriately set the specified value of the certainty factor. Therefore, in the present embodiment, it is preferable to adjust the prescribed values of the certainty factor and the reference value P according to the density of the parts scrap in the image so that the recognition error rate is 5% or less. For example, when the parts density of electronic and electrical equipment parts scrap is high and there is a lot of overlap between parts scraps, it is preferable to set the reference value P to 0.9 or more, the parts density is low, and the gap is more than a certain level It is preferable to set the reference value P to 0.7 or more when the scrap parts are arranged with each other.

Based on the classification data stored in the classification data storage means 111, the classification means 101 assigns a recognition frame to the image containing the electronic/electrical equipment parts scrap after a predetermined correction process, and then draws the image under different drawing conditions. to classify each of a plurality of scrap parts. For example, when it is desired to classify a screen on which electronic/electrical device component waste is imaged into three types of board, plastic, and other component waste, the classification means 101 classifies the board, plastic, and other component waste into electronic/electrical device Different types of target objects are extracted from the picked-up images of the component wastes, and recognition frames with different drawing conditions are assigned to the extracted substrates, plastics, and other component wastes. FIG. 3 shows an example of an image in which the sorting means 101 specifies electronic/electrical device parts scraps existing in the image for each parts scrap and classifies each parts scrap with a recognition frame of a different color.

The measuring means 102 measures the area of each image of a plurality of scrap parts classified by the classifying means. As shown in FIG. 3, the measuring means 102 measures the areas of the scrap parts separated by different color frames using the measurement information stored in the analysis information storage means 112 for each piece of scrap parts. The analysis information storage means 112 stores information necessary for analysis of the composition analysis by the processing device 100, such as information for measuring the area of a plurality of scrap parts, information for each scrap scrap required for calculating the weight ratio of scrap scrap parts. Information such as the weight ratio per unit is stored.

The analyzing means 103 estimates the composition of the electronic/electrical device scrap by multiplying the estimated weight per unit area of the plurality of scrap parts by the area of the plurality of scrap parts and analyzing the weight ratio of the plurality of scrap parts. do. The estimated weight per unit area of a plurality of scrap parts can be set in advance by the operator via the input device 120 or the like according to the operation result. Although not limited to the following, for example, when electronic and electrical equipment parts scraps are classified into three types, the assumed weight of substrate scraps is, for example, 2.0 g/cm ² , and the assumed weight of plastics is 1.5 g/cm. ² , and other parts can be set at 1.0 g/cm ² .

In addition to the area, the analysis means 103 analyzes the number (pieces) of the parts constituting the parts scrap for each scrap, the average particle diameter calculated based on the above calculation result of the area and the number, the weight ratio, and the like. Physical characteristics can also be analyzed and output to the display device 130 or the like. An example of analysis results is shown in FIG. The analysis results shown in FIG. 3 show that the substrate accounts for 70% of the total, followed by plastics at 22%, copper wire scraps at 6%, and metal chips at 3%.

The operating condition generating means 104 generates information on the operating conditions of a sorter for sorting out a plurality of scrap parts based on the analysis results of the weight ratios of the plurality of scrap parts by the analysis means. As the sorting machine, there are various sorting machines such as picking, color sorter, metal sorter, eddy current sorter, wind sorter, and sieving machine. For example, from the analysis results shown in FIG. 4, the operating condition generating means 104 generates operating conditions for a color sorter that sorts substrates and plastics, and stores the generated operating conditions in the operating condition storage means 113 . The operating conditions stored in the operating condition storage means 113 are output to the sorters 13 and 14, and the sorters 13 and 14 can perform sorting processing according to the output operating conditions. The change information generation means 105 stores the change conditions of the operating conditions in the operating condition storage means 113 .

The position information output means 106 acquires the position information of each of the plurality of scrap parts classified by the classification means 101 in the image of electronic/equipment scrap parts, and stores the information in the position information storage means 114 . Then, the positional information is output to specific sorters 13 and 14 for extracting the position of a specific piece of scrap from among a plurality of pieces of scrap and sorting it out. For example, substrates and metal pieces cannot be separated by specific sorters 13 and 14 such as metal sorters, but if individual position information can be obtained from image information, they can be sorted by sorters 13 and 14 having a picking function. will be able to

The machine learning means 107 mechanically classifies the classification data referred to by the classification means 101 based on the selection results of the sorters 13 and 14 for sorting out a plurality of images of electronic/electrical device parts scraps or a plurality of parts scraps. Processing by learning.

In machine learning, first, a recognition frame is attached to a single component scrap that appears in an image of electronic/electrical equipment component scrap. For example, a recognition frame is attached to the board in the image by encircling it with a red line using paint software, and the machine learning means 107 further learns that the component waste marked with the recognition frame is the board. . Similarly, the machine learning means 107 learns that plastics, metal pieces, copper wire scraps, and the like surrounded by recognition frames of several kinds of colors are each unique parts scraps. Learning data of 100 or more sheets, preferably several hundred sheets to several thousand sheets are prepared by such a method, and machine learning means 107 processes the learning data to learn the characteristics of each scrap and classify it. learn to improve the accuracy of

Based on the learning result of the machine learning means 107, the updating means 108 updates the classification data used by the classification means 101 to classify the scrap parts. The updated classification data may be output to the sorting machine 14 and server 15 connected via the network 11 .

According to the processing apparatus for electronic and electrical equipment component scraps according to the embodiment of the present invention, the recognition frame providing means 141 provided in the image extracting means 140 captures images of a plurality of objects O1 and O2 having different shapes or areas. , based on the classification data containing the recognition information of the objects O1 and O2, the correction means 142 applies recognition frames 1 to 6 to the object O1 , O2 are extracted from the image, an inappropriate recognition frame is deleted from arbitrary two recognition frames 1 to 6 included in the extracted objects O1 and O2, and the actual objects O1 and O2 are extracted. By correcting the number of recognition frames 1 to 6 to match the number of , it is possible to reduce erroneous recognition due to image recognition and improve recognition accuracy. This makes it possible to perform a more realistic and appropriate analysis, for example, when analyzing the existence ratio and composition of the objects O1 and O2 using the recognition frames attached to the images.

(Object image extraction processing method)
An example of the object image extraction processing method according to the embodiment of the present invention will be described with reference to the flowchart shown in FIG. In step S<b>1 , the recognition frame providing means 141 provided in the image extracting means 140 gives a recognition frame to the object based on the classification data containing the object recognition information stored in the classification data storage means 111 . The result of giving the recognition frame is stored in the storage device 110 .

In step S2, the correcting means 142 included in the image extracting means 140 extracts objects having two or more overlapping recognition frames from the images to which the recognition frames have been added by the recognition frame adding means 141. FIG. In step S3, the correcting means 142 determines whether or not three or more recognition frames of the extracted object overlap. If three or more recognition frames overlap, the process proceeds to step S4, two large and small recognition frames that overlap with each other are extracted, and the process proceeds to step S5. If the number of recognition frames assigned to the extracted target object is two, the process directly proceeds to step S5.

In step S5, the correcting means 142 calculates the area S _o of the overlapping portion of the two overlapping large and small recognition frames, and the area S _s of the recognition frame 2, which has the smaller area among the two recognition frames 1 and 2. . The calculation result is stored in the storage device 110 as appropriate. The correcting means 142 further adjusts the ratio of the area S _o of the overlapping portion to the area S _s of the smaller one of the two recognition frames (area ratio S _o /S _s ) by a preset reference value P Compare with If the area ratio S _o /S _s <reference value P, the process proceeds to step S8. If the area ratio S _o /S _s ≥ the reference value P, the process proceeds to step S7.

In step S7, the correction unit 142 deletes the recognition frame with the smaller area among the two recognition frames. In step S8, it is determined whether or not there is another target object having two or more overlapping recognition frames to be extracted, and if there is another target object to be extracted, the process returns to step S2. If there are no other objects, the process ends.

According to the object image extraction processing method according to the embodiment of the present invention, even if a plurality of recognition frames are attached due to erroneous recognition when the object is one, the number of recognition frames can be appropriately determined. Since correction is possible, it is possible to provide an image extraction processing method capable of further improving recognition accuracy.

(Image extraction processing and composition analysis method of object using electronic and electrical equipment parts scrap)
An example of an image extraction processing method, a composition analysis method, and a processing method for electronic/electrical device parts scrap using the electronic/electrical device parts scrap processing apparatus shown in FIG. 1 will be described with reference to the flowchart of FIG. . Note that steps S11 to S18 in FIG. 6 are substantially the same as steps S1 to S8 in FIG. 5, and redundant descriptions are omitted.

At step S19, the classification means 101 of FIG. , plastics, metal chips, copper wire scraps, capacitors, IC chips, and other parts scraps).

The classification result by the classification means 101 can be displayed by the display device 130 or the like. The classification results are displayed with different drawing conditions, such as color, frame thickness, and frame line (dotted line, broken line, double line, etc.), so that the operator can easily confirm. In the image displayed on the display device 130, for example, a recognition frame with a different color is attached to each piece of scrap, such as a red frame for a board and a blue frame for plastic. At this time, the positional information output means in FIG.

For example, when classified into 7 categories: substrates, plastics, metal scraps, copper wire scraps, capacitors, IC chips, and other component scraps, substrates, copper wire scraps, capacitors, and IC chips are classified as valuables, and metal scraps (aluminum, By optimizing the sorting conditions so that SUS) and plastics are sorted out as smelting inhibitors, it is possible to quantify and manage the separation efficiency, loss rate, and operation results of electronic and electrical equipment parts scraps. .

Furthermore, in step S19, the measuring means 102 measures the total area of the plurality of scrap parts sorted by the sorting means 101. FIG. The area of each scrap component can be measured by the measuring means 102 using known area detection software for estimating the area from the image. In step S20, the analysis means 103 multiplies the estimated weight per unit area of the plurality of scrap parts by the area of the plurality of scrap parts, and analyzes the weight ratio of the plurality of scrap parts to obtain the electronic/electrical equipment parts scrap. Estimate the composition of

For example, as shown in FIG. 3, the analysis means 103 can digitize and analyze the average area, number, average particle diameter, weight ratio, etc. of each of a plurality of scrap parts from the image captured by the imaging device 12. Therefore, the raw material composition can be quantified and grasped much more quickly than the conventional method of evaluating the raw material composition of electronic/electrical device parts scrap by hand.

In step S21, sorting processing is performed based on the raw material analysis result analyzed by the analysis means 103. FIG. For example, based on the composition analysis result obtained in step S20, the selection of a sorter for sorting the raw material, and the operating conditions such as sorting conditions and sorting order are determined, and the sorting process is performed based on the operating conditions. done.

In this way, by utilizing the composition analysis results obtained by the composition analysis apparatus according to the embodiment of the present invention for each of the sorting machines 13, while continuously photographing electronic and electrical equipment component scraps being transported, Image data can be analyzed in real time to analyze raw material composition.

Conventionally, the raw material composition of electronic and electrical equipment parts scrap was evaluated by manual sorting, and the results were reflected in the operation management of the sorting process and the setting of operating conditions. In the method of grasping , it was not possible to perform rapid processing.

According to the embodiment of the present invention, the composition of electronic/electrical device parts whose composition changes from moment to moment is instantaneously determined by image analysis and separation based on predetermined classification data. Since it can be quantified, it is possible to quickly sort a large amount of electronic and electrical equipment parts scrap under more appropriate conditions.

In addition, since the analysis results are displayed on the display device 130 with a frame of a different color for each raw material type, the operator can easily recognize the analysis results, so that erroneous detection by the composition analysis apparatus can be easily recognized.

Further, the recognition frame assigned to each scrap component is corrected and optimized in each process shown in steps S12 to S18, thereby reducing erroneous recognition in image recognition and improving analysis accuracy. can be improved.

Although the present invention has been described by the above embodiments, the statements and drawings forming part of this disclosure should not be understood to limit the present invention. That is, the present invention is not limited to each embodiment, and can be embodied by modifying the constituent elements without departing from the spirit of the present invention. Moreover, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in each embodiment. For example, some components may be deleted from all components shown in the embodiments. Furthermore, components of different embodiments may be combined as appropriate.

Examples of the present invention are presented below along with comparative examples, which are provided for a better understanding of the invention and its advantages and are not intended to be limiting of the invention.

(Example 1)
Targeting electronic and electrical equipment parts scraps, recognition frames were added to images of parts scraps composed of a plurality of substrate scraps taken by the processing apparatus according to the present embodiment, and objects with two or more recognition frames overlapping were extracted. It was found that objects with two or more overlapping recognition frames can be classified into the following three cases.

(1) When the small frame B is completely inside the large frame A (when the area is frame A ∩ frame B/frame B = 100%)
(2) When the small frame B is inside the large frame A and has a reference value (90% here) or more and less than 100% (the area is 90% ≤ frame A ∩ frame B / frame B < 100% case)
(3) When two frames partially overlap (when the area is frame A ∩ frame B/frame B < 90%)

As a result of examining the result of classifying substrate scraps in which two or more recognition frames overlapped in the image in the above-described modes (1) to (3), the proportion of case (1) was 9.8%, and that of case (2). accounts for 58.5%, and cases (1) and (2) account for 68.3% in total. Also, the actual number N _o of substrate scraps in the image is visually counted, and the ratio (N _o /N _c ) to the recognition frame N _c given by the image recognition processing of the processing device. The misrecognition rate was 13.8%.

After that, when the recognition frame in which 90% or more of the small frame B is inside the large frame A (recognition frame corresponding to the above-described cases (1) and (2)) is deleted by the correction processing according to the present embodiment, The erroneous recognition rate of the object represented by the ratio (N _o /N _c ) to the recognition frame N _c given by the image recognition processing of the processing device was able to be reduced to 4.4%.

(Example 2)
Targeting electronic and electrical equipment parts scraps, the processing apparatus according to this embodiment attaches a recognition frame to an image of a plurality of plastic parts scraps captured, extracts objects with two or more recognition frames overlapping, and implements. When classified into the same three cases as in Example 1, the proportion of case (1) was 26.7%, and the proportion of case (2) was 43.3%, for a total of 70 cases (1) and (2). .0%. Also, the actual number N _o of substrate scraps in the image is visually counted, and the ratio (N _o /N _c ) to the recognition frame N _c given by the image recognition processing of the processing device. The misrecognition rate was 12.3%.

After that, when the recognition frame in which 90% or more of the small frame B is inside the large frame A (recognition frame corresponding to the above-described cases (1) and (2)) is deleted by the correction processing according to the present embodiment, The erroneous recognition rate of the object represented by the ratio (N _o /N _c ) to the recognition frame N _c given by the image recognition processing of the processing device was able to be reduced to 3.7%.

As described above, according to the processing method according to the present embodiment, the number of recognition frames given to an object by image recognition can be brought closer to the number of actual objects, thereby improving the recognition accuracy of image recognition. becomes possible.

DESCRIPTION OF SYMBOLS 10... Composition-analysis apparatus 11... Network 12... Imaging apparatus 13, 14... Sorter 15... Server 100... Processing apparatus 101... Classification means 102... Measurement means 103... Analysis means 104... Operating condition generation means 105... Change information generation means 106 Position information output means 107 Machine learning means 108 Update means 110 Storage device 111 Classification data storage means 112 Analysis information storage means 113 Operating condition storage means 114 Position information storage means 120 Input device 130 Display device 140... Image extracting means 141... Recognition frame adding means 142... Correcting means

Claims (8)

Giving a recognition frame with a different drawing condition to each of the plurality of scrap parts from an image obtained by imaging electronic/electrical device scrap including a plurality of scrap parts having different shapes or areas,
Extracting from the image the scrap parts in which two or more of the recognition frames overlap, calculating the area of the overlapping portion of any two recognition frames included in the extracted scrap parts, deleting the smaller recognition frame when the ratio of the area of the overlapped portion to the area of the recognition frame having the smaller area is equal to or greater than a reference value,
classifying a plurality of scrap parts to which the recognition frame is attached, which are included in the image after deletion, for each of the plurality of scrap scrap parts and measuring the total area of each of the scrap scrap parts;
Estimating the composition of the electronic/electrical device scrap by multiplying the total area of the plurality of scrap parts by the assumed weight per unit area of the scrap parts and analyzing the weight ratio of the scrap parts A composition analysis method for electronic/electrical equipment parts scrap, comprising: 2. The method for analyzing the composition of scrap electronic/electrical equipment parts according to claim 1, further comprising repeating the process of deleting the smaller recognition frame when three or more of the recognition frames overlap until there are no more recognition frames to be deleted. . Giving a recognition frame to each of the plurality of scrap parts is a minimum circumscribing figure that circumscribes the scrap parts based on classification data including recognition information including components and colors of the scrap parts. 3. The method for analyzing the composition of electronic/electrical equipment component scraps according to claim 1 or 2, comprising giving a . 4. The method for analyzing the composition of electronic and electrical equipment component scraps according to any one of claims 1 to 3, wherein the reference value is 0.75 to 1.0. Extracting the component scraps from the image includes classifying the electronic/electrical device component scraps into substrates, plastics, and other component scraps, and extracting the extracted substrates and plastics. 5. The composition analysis method for electronic/electrical equipment parts scrap according to any one of claims 1 to 4, further comprising giving recognition frames under different drawing conditions to each of the other parts scrap and the other parts scrap. Extracting the scrap parts from the image includes classifying and extracting the scrap electronic/electrical equipment parts into substrates, plastics, metal pieces, copper wire scraps, capacitors, IC chips, and other scrap scraps. and providing a recognition frame under different drawing conditions to each of the extracted substrate, plastic, metal piece, copper wire scrap, capacitor, IC chip, and other component scrap. 5. The method for analyzing the composition of electronic/electrical equipment parts scrap according to any one of 1 to 4. Recognition information of the scrap parts for assigning a recognition frame with a different drawing condition to each of the scrap parts from an image of electronic/electrical equipment scrap including a plurality of scrap parts having different shapes or areas. a storage device comprising classification data including
a recognition frame providing means for providing a recognition frame to the plurality of scrap parts based on the classification data;
Extracting from the image the scrap parts in which two or more of the recognition frames overlap, calculating the area of the overlapping portion of any two recognition frames included in the extracted scrap parts, Correction for correcting the recognition frame by deleting the smaller recognition frame when the ratio of the area of the overlapping portion to the area of the recognition frame having the smaller area among the recognition frames is equal to or greater than a reference value. means and
measuring means for classifying a plurality of scrap parts to which the recognition frame is attached and included in the image after deletion and measuring the total area of each of the plurality of scrap scrap parts;
Estimating the composition of the electronic/electrical device scrap by multiplying the total area of the plurality of scrap parts by the assumed weight per unit area of the scrap parts and analyzing the weight ratio of the scrap parts and an analysis means for analyzing the composition of electronic/electrical equipment component scraps. An imaging step of imaging electronic/electrical device parts scrap consisting of a plurality of parts scraps;
Based on the classification data including the recognition information of the component scraps for assigning a recognition frame with a different drawing condition to each of the plurality of component scraps from the image of the electronic / electrical equipment component scraps, a recognition frame assigning step of extracting images of scrap parts and assigning a recognition frame to each of the plurality of scrap parts;
Extracting from the image the scrap parts in which two or more of the recognition frames overlap, calculating the area of the overlapping portion of any two recognition frames included in the extracted scrap parts, Correction for correcting the recognition frame by deleting the smaller recognition frame when the ratio of the area of the overlapping portion to the area of the recognition frame having the smaller area among the recognition frames is equal to or greater than a reference value. process and
a measuring step of classifying a plurality of scrap parts to which the recognition frame is attached and included in the image after the correction process and measuring the total area of each of the plurality of scrap scrap parts;
Analyzing the composition of the electronic/electrical device scrap by multiplying the estimated weight per unit area of the plurality of scrap parts by the total area of the plurality of scrap parts and analyzing the weight ratio of the plurality of scrap parts. an analysis step to
A method for processing electronic and electrical equipment parts scraps, comprising:

Images