JP2004086799A - Method and device for individually recognizing object - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and a device for individually recognizing objects capable of individually recognizing individual objects in a state that a plurality of objects are congested and overlapped even without using an aligning device and a segmenting device for aligning and unraveling the objects. <P>SOLUTION: A plurality of objects 20 on a shaking table 1 are individually and minutely moved by a shaker 2, images photographed at intervals with a camera 3 during minute movement are inputted into an image processor 4 and the existence and positions of the individual objects are recognized on the basis of changes of images photographed at intervals. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an individual recognition method and an individual recognition device for individually recognizing each object in a state where a plurality of objects to be processed overlap each other, such as a waste sorting line.
[0002]
[Prior art]
Recycling-based social systems have been improved, various types of waste recycling facilities have been constructed, and sorting efficiency has been improved by mechanizing or automating conventional manual sorting operations.
[0003]
Among them, for the operation of identifying the material of the waste, many methods have been proposed for identifying the material based on optical information such as visible light, infrared light, and ultraviolet light.
[0004]
For example, Japanese Patent Application Laid-Open No. 9-89768 discloses a method of irradiating near-infrared rays to a waste plastic bottle which is conveyed while being aligned on a conveyor, and discriminating a plastic material from its absorbance. Also, Japanese Patent Application Laid-Open No. Hei 10-232166 discloses a method of irradiating a paper pack on a conveyor with near infrared rays and identifying the material of the paper pack from its absorbance.
[0005]
In these material identification methods, it is indispensable to align the identification objects in a dense or overlapping state one by one at predetermined intervals so as to face the identification sensor. Many inventions relating to the device have been made. For example, JP-A-8-143138 and JP-A-2002-114363 disclose an aligning apparatus provided with a dedicated conveyor for alignment.
[0006]
[Problems to be solved by the invention]
However, these alignment devices have the following problems.
[0007]
First, the accuracy of arranging the objects one by one at predetermined intervals is not sufficient, and sometimes a plurality of objects may be recognized as one object. For this reason, accurate identification cannot be performed in the subsequent material identification step, which causes a reduction in identification accuracy.
[0008]
In addition, the increase in the number of dedicated conveyors and cutting-out mechanisms for alignment increases the cost of the entire equipment, and requires space for installing them.
[0009]
The present invention has been made to solve the above problems, and a plurality of objects are densely or overlapped without using an alignment device or a cutting device for aligning or loosening the objects. It is an object of the present invention to provide an individual recognition method and an individual recognition device for an object capable of individually recognizing each target object in a state where the objects are kept in a state.
[0010]
[Means for Solving the Problems]
In order to solve the above problems, the present invention has the following features.
[0011]
[1] An individual recognition method for individually recognizing a plurality of objects, in which vibrations are simultaneously applied to the plurality of objects, and the plurality of objects being vibrated are photographed at intervals. A method for individually recognizing an object, comprising determining a speed vector for each pixel forming an image from an image captured at a time interval, and recognizing a group of pixels having the same speed vector as one object.
[0012]
[2] An individual recognizing device for individually recognizing a plurality of objects, a vibrating means for simultaneously vibrating the plurality of objects, and a plurality of objects vibrated by the vibrating means. And an image processing unit that determines a velocity vector for each pixel forming an image from an image captured by the imaging unit at a time interval, and recognizes a group of pixels having the same velocity vector as one object. An individual recognition device for an object, comprising:
[0013]
BEST MODE FOR CARRYING OUT THE INVENTION
(1st Embodiment)
FIG. 1 shows a first embodiment of the present invention.
[0014]
The individual object recognition device according to the first embodiment of the present invention includes a vibration table 1, a camera 3, and an image processing device 4.
[0015]
The vibration table 1 includes a table 1a and a vibrator 2 for applying vibration to the table 1a, and is configured to vibrate by placing a plurality of objects 20 on an upper surface thereof.
[0016]
The camera 3 is provided above the vibration table 1, photographs an object 20 on the vibration table 1, and sends the image to the image processing device 4.
[0017]
The image processing device 4 captures an image captured by the camera 3 and performs a predetermined process.
[0018]
Next, a method of individually recognizing a plurality of objects 20 in a dense or overlapping state on the vibration table 1 using the individual recognition device configured as described above will be described with reference to a processing flowchart shown in FIG. This will be described with reference to FIG.
[0019]
First, vibration is applied to the vibration table 1 by the vibrator 2. As a result, the plurality of objects 20 densely or overlapping on the vibration table 1 are individually minutely moved.
[0020]
Then, as shown in FIG. 2A, at an arbitrary time t = t during the minute movement, a plurality of objects 20 are photographed by the camera 3, and the images are used as an image matrix M (t). Take in 4.
[0021]
Next, as shown in FIG. 2B, at time t = t + Δt after a lapse of a short time Δt from time t, a plurality of objects 20 are photographed again by the camera 3, and the images are converted into an image matrix M ( (t + Δt) is taken into the image processing apparatus 4.
[0022]
Next, as shown in FIG. 2C, a predetermined function F is used to move each pixel forming an image for a time period Δt from the image matrix M (t) and the image matrix M (t + Δt). The quantity, that is, the velocity vector Vi (t) for each pixel is obtained.
[0023]
Then, as shown in FIG. 2D, a speed vector matrix W (t) having the speed vector Vi (t) for each pixel as an element is defined and stored in the image processing device 4. That is, W (t) = W (Vi (t)).
[0024]
Next, as shown in FIG. 2E, pixels having the same speed vector Vi (t) are grouped in the speed vector matrix W (t).
[0025]
As a result, as shown in FIG. 2F, the entire image is divided into a plurality of regions for each group of pixels.
[0026]
Then, the plurality of regions obtained as described above are each recognized as an individual object.
[0027]
That is, since the speed vectors of the same object are the same, but the speed vectors of the adjacent objects are different, it is possible to recognize that the boundary line at which the speed vector changes is the boundary of each object.
[0028]
FIG. 3 shows a specific example of the above procedure.
[0029]
First, as shown in FIG. 3A, three objects A, B, and C are overlapped on the vibration table 1, and the objects A, B, and It is assumed that C has slightly moved in the directions as shown in FIG.
[0030]
Then, from the calculation of the image matrix M (t) obtained at the time t = t and the image matrix M (t + Δt) obtained at the time t = t + Δt, as shown in FIG. The vector Vi (t) is obtained. The speed vector matrix W (t) stores the speed vector Vi (t) for each pixel.
[0031]
By grouping pixels having the same speed vector Vi (t) in the speed vector matrix W (t), three pixels corresponding to the objects A, B and C are grouped as shown in FIG. It is divided into areas.
[0032]
Thereby, the existence and the position of the objects A, B, and C can be individually recognized.
[0033]
It should be noted that whether or not the speed vectors Vi (t) of the pixels are equal in the above is determined in advance according to the difference between the size and the direction of the speed vector Vi (t) between the pixels according to the object to be recognized. The determination is made by judging that the speed vectors Vi (t) of the pixels are equal to each other when the threshold value is within the predetermined threshold value, and by judging that the speed vectors Vi (t) of the pixels are not equal to each other when the threshold value is exceeded.
[0034]
In this manner, in this embodiment, the existence and position of each individual object can be individually recognized even when a plurality of objects are densely or overlapped. Therefore, in the subsequent processing steps, it is possible to accurately perform material identification and the like of each target object based on the recognition information, and it is not necessary to provide a special alignment device or cutting device for material identification or the like. .
(Second embodiment)
A waste sorting apparatus to which the individual object recognition apparatus according to the present invention is applied is shown in FIG. 4 as a second embodiment of the present invention.
[0035]
The waste sorting device according to the second embodiment includes a conveyor 10, a vibrator 2, a camera 3, an image processing device 4, a sensor unit 5, an identification device 6, a handling robot 7, a robot controller 8, a control device 9, The sorting chute 11 comprises a sorting chute 11 and sorts a large number of wastes 21 conveyed by the conveyor 10 into predetermined sorting chute 11 for each type.
[0036]
The conveyor 10 moves with a large number of wastes 21 to be sorted placed thereon.
[0037]
The vibrator 2 is attached to the lower part of the upstream part of the conveyor 10, and applies a predetermined vibration by hitting the conveyor 10 from the back surface.
[0038]
The camera 3 is provided above the conveyor 10 slightly downstream of the mounting position of the vibration exciter 2, photographs the waste 21 on the conveyor 10, and sends the image to the image processing device 4.
[0039]
The image processing device 4 captures an image captured by the camera 3, performs a predetermined process, and individually recognizes the waste 21 on the conveyor 10.
[0040]
The sensor unit 5 is provided so as to straddle the conveyor 10 on the downstream side from the installation position of the camera 3, and has a predetermined shape necessary for identifying the waste 21 on the conveyor 10 passing through the sensor unit 5. An identification sensor is provided. For example, a material identification sensor for identifying a material, a color identification sensor for a color, or the like is used.
[0041]
The identification device 6 identifies the type of the waste 21 on the conveyor 10 based on the output from the sensor unit 5.
[0042]
The handling robot 7 is provided on the side of the conveyor 10 on the downstream side from the installation position of the sensor unit 5, grasps a predetermined waste from the waste 21 on the conveyor 10, and puts it into the corresponding sorting chute 11. I do.
[0043]
The robot controller 8 instructs the handling robot 7 to operate.
[0044]
The control device 9 integrates the image processing device 4, the identification device 6, and the robot controller 8, and operates the handling robot 7 through the robot controller 8 based on the recognition information from the image processing device 4 and the identification information from the identification device 6. Control.
[0045]
In the waste sorting apparatus configured as described above, a procedure for sorting a large number of wastes 21 conveyed in a state of being densely or overlapping on the conveyor 10 into predetermined sorting chutes 11 for each type will be described below. Will be described.
[0046]
First, the conveyer 10 is vibrated by the vibrator 2 to pass the waste 21 on the conveyor 10 under the camera 3 while slightly changing the relative position between the wastes. Then, the image on the conveyor 10 taken by the camera 3 is sent to the image processing device 4.
[0047]
Then, the image processing apparatus 4 calculates the speed of each pixel from the image matrix M (t) and the image matrix M (t + Δt) obtained at intervals of the short time Δt by the same processing flow as described in the first embodiment. By calculating the vector Vi (t) and grouping pixels having the same speed vector Vi (t), the position of each waste 21 is individually recognized, and the recognition information is sent to the control device 9.
[0048]
Then, assuming that the distance between the camera 3 and the sensor unit 5 in the direction of travel of the conveyor is x (m) and the speed of the conveyor 10 is v (m / s), the waste 21 that has passed through the camera 3 is τ = x / v (S) Later, it is sensed by the identification sensor, and based on the identification signal, the identification device 6 identifies the type of the waste and sends the identification information to the control device 9.
[0049]
The control device 9 takes into account the time delay τ and, based on the recognition information from the image processing device 4 and the identification information from the identification device 6, sends information on the position and type of the target waste 21 to the robot controller. 8
[0050]
The robot controller 8 instructs the handling robot 7 to operate based on the information from the control device 9, and the handling robot 7 grasps the target waste 21 from the conveyor 10 based on the instruction and performs predetermined sorting. Put it in the chute 11.
[0051]
For example, when sorting empty bins conveyed by the conveyor 10 into transparent bins and colored bins, a color identification sensor is attached as an identification sensor, and the handling robot 7 operates based on the empty bin position information and color information. The transparent bottle is grasped from the conveyor 10 and thrown into the transparent bottle chute, and the colored bottle is grasped and thrown into the colored bin chute.
[0052]
In this manner, in this embodiment, a large number of wastes stacked and conveyed on the conveyor 10 are accurately sorted into predetermined sorting chutes 11 for each type without using an aligning device or a cutting device. be able to.
[0053]
(Third embodiment)
FIG. 5 shows a waste plastic sorting apparatus to which the individual object recognition apparatus according to the present invention is applied, as a third embodiment of the present invention.
[0054]
The waste plastic sorting device according to the third embodiment includes a roller conveyor 10a, a container 12, a shaker 2, a camera 3, an image processing device 4, a material identification sensor 5a, an identification device 6, a handling robot 7, and a robot controller 8. , A control device 9 and a sorting chute 11. The waste plastic 22 contained in the container 12 conveyed by the roller conveyor 10 a is directly taken out of the container 12, and a predetermined sorting chute is set for each material. This is a device for sorting into 11 units.
[0055]
The roller conveyor 10a is configured to move with the container 12 placed on its upper surface and stop at a predetermined position.
[0056]
The container 12 is open at the top, and accommodates a large number of various waste plastics 22 in a stack.
[0057]
The vibrator 2 is installed at a position below the roller conveyor 10a, and hits the outer surface of the target container 12a stopped at a predetermined position above the vibrator 2 from a gap between the rollers of the roller conveyor 10a. A predetermined vibration is applied to the target container 12a.
[0058]
The camera 3 is provided above the stop position of the target container 12a, photographs the waste plastic 22 in the target container 12 from above, and sends the image to the image processing device 4.
[0059]
The image processing device 4 captures the image captured by the camera 3, performs a predetermined process, and individually recognizes the waste plastic 22 in the target container 12a.
[0060]
The material identification sensor 5a is provided between the stop position of the target container 12a and the installation position of the sorting chute 11, and is a sensor in which an infrared irradiation device and a detector that detects reflected light thereof are integrated.
[0061]
The identification device 6 uses the fact that there is a difference in the infrared absorptance depending on the material of the waste plastic, and determines the material of the waste plastic based on the reflection spectrum of the infrared ray detected by the material identification sensor 5a, as PS, PE, PP, or PVC. , PET, ABS, and the like.
[0062]
The handling robot 7 is provided in the vicinity of the stop position of the target container 12a, grasps the waste plastic in the target container 12a, passes the waste plastic through the field of view of the sensor 5a, and places the waste plastic in a predetermined sorting chute 11. throw into.
[0063]
The robot controller 8 instructs the handling robot 7 to operate.
[0064]
The control device 9 integrates the image processing device 4, the identification device 6, and the robot controller 8, and operates the handling robot 7 through the robot controller 8 based on the recognition information from the image processing device 4 and the identification information from the identification device 6. Control.
[0065]
A procedure for sorting the waste plastics 22 stored in the container 12 in a stacked state in the waste plastic sorting apparatus configured as described above into predetermined sorting chutes 11 for each material will be described below.
[0066]
First, the target container 12a on the roller conveyor 10a is stopped at a predetermined position above the vibrator 2, and a predetermined vibration is applied to the target container 12a by hitting the outer surface of the target container 12a by the vibrator 2. Accordingly, the waste plastic 22 in the target container 12a slightly changes the relative position between the waste plastics.
[0067]
Then, the image inside the target container 12 a taken from above by the camera 3 is sent to the image processing device 4.
[0068]
Then, the image processing apparatus 4 calculates the speed of each pixel from the image matrix M (t) and the image matrix M (t + Δt) obtained at intervals of the short time Δt by the same processing flow as described in the first embodiment. By calculating the vector Vi (t) and grouping pixels having the same speed vector Vi (t), the position of each waste plastic 22 is individually recognized, and the recognition information is sent to the control device 9.
[0069]
Then, the control device 9 transmits the position information of the target waste plastic 22 to the robot controller 8 based on the recognition information.
[0070]
Then, the robot controller 8 instructs the handling robot 7 to operate based on the position information from the control device 9, and the handling robot 7 grasps the target waste plastic 22 from the target container 12a based on the instruction. Then, the waste plastic 22 passes through the visual field of the material identification sensor 5a.
[0071]
Then, based on the reflection spectrum of the infrared ray detected by the material identification sensor 5a, the identification device 6 identifies whether the material of the waste plastic 22 is PS, PE, PP, PVC, PET, ABS, or the like.
[0072]
Then, the material information is sent to the control device 9, and the control device 9 specifies the sorting chute 11 corresponding to the material information to the robot controller 8.
[0073]
Then, the robot controller 8 instructs the handling robot 7 to put the waste plastic 22 into the sorting chute 11 corresponding to the material information.
[0074]
Hereinafter, the above-described procedure is repeated, and the waste plastics 22 in the target container 12a are sequentially sorted, and the process is continued until the target container 12a becomes empty. Whether or not the target container 12a is empty can be determined by an image from the camera 3, but more simply, the weight of the target container 12a is monitored by a load cell or the like, and the weight of the target container 12a is determined by a predetermined threshold value. It may be determined that it is empty when it becomes less than.
[0075]
In this manner, in this embodiment, the waste plastics 22 stored in a stacked state in the container 12 can be accurately placed on the predetermined sorting chute 11 for each material without using an aligning device or a cutting device. Can be sorted.
[0076]
In this embodiment, the material selection of the waste plastic is performed. However, it is also possible to perform the color selection of empty bottles by using a color identification sensor instead of the material identification sensor 5a.
[0077]
In addition, it is also possible to prepare various types of sensors in advance, and to sort different types of waste by the same sorting device, for example, every day of the week or every hour.
[0078]
【The invention's effect】
ADVANTAGE OF THE INVENTION According to this invention, the presence and position of each object can be individually recognized even in the state where a plurality of objects are densely or overlapped.
[0079]
Therefore, it is not necessary to provide an alignment device or a cutting device for aligning or loosening the target object, and the entire processing apparatus can be reduced in size.
[Brief description of the drawings]
FIG. 1 is an explanatory diagram of a first embodiment according to the present invention.
FIG. 2 is a processing flowchart according to the first embodiment of the present invention.
FIG. 3 is an explanatory diagram of an image processing procedure according to the first embodiment of the present invention.
FIG. 4 is an explanatory diagram of a second embodiment according to the present invention.
FIG. 5 is an explanatory diagram of a third embodiment according to the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Vibration table 1a Table 2 Exciter 3 Camera 4 Image processing device 5 Sensor unit 5a Material identification sensor 6 Identification device 7 Handling robot 8 Robot controller 9 Control device 10 Conveyor 10a Roller conveyor 11 Sorting chute 12 Container 12a Target container 20 Target object 21 Waste 22 Waste plastic

Claims (2)

An individual recognition method for individually recognizing a plurality of objects, in which vibrations are simultaneously applied to the plurality of objects, and the plurality of objects being vibrated are photographed with a time interval therebetween. A speed vector for each pixel forming an image from an image photographed by the camera, and recognizing a group of pixels having the same speed vector as one object. An individual recognizing device for individually recognizing a plurality of objects, comprising: vibrating means for simultaneously vibrating a plurality of objects; and photographing the plurality of objects vibrated by the vibrating means. An imaging unit, and an image processing unit that obtains a speed vector for each pixel forming an image from an image captured by the imaging unit at a time interval and recognizes a group of pixels having the same speed vector as one object. Individual recognition device for featured objects.

Images