JPH07286969A - Discriminating method for copper including scraps - Google Patents

Abstract

PURPOSE:To automatically discriminate the copper including scraps in an iron scrap group at a high accuracy, by finding the brightness value, the saturation value, and the hue angle value, from the RGB signals the picture elements in the pickup images of the iron scrap group have, and comparing them with specific values. CONSTITUTION:An RGB signal 4 obtained by photographing the scope to require the copper discrimination in an iron scrap group 1 by a TV camera 3 is converted into signals of the hue angle 6, the saturation 7, and the brightness 8, by an HSI converter 5, and they are delivered to a discrimination processing device 9. The discrimination processing device 9 finds the hue angles 6 of the picture elements, when the saturations 7 of the picture elements are more than the threshold value and compares with and refers to the corresponding relation of the hue angle scope of the copper, and when the hue angle 6 is within the specific scope, the picture elements are decided to be the copper. When the hue angle 6 is not within the scope, it is decided not to be the copper, and when the saturation 8 is less than a threshold value, it is decided not to be the copper. And when the ratio of the total area of individual scraps, and the copper area in the image area the scraps occupy is a threshold value or higher, it is decided to be the copper including scraps.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for automatically identifying scrap containing copper, which is an impurity element, from a group of iron scraps in iron scrap regenerative processing.

[0002]

2. Description of the Related Art In order to avoid deterioration of the quality of iron produced by scrap regeneration, it is necessary to prevent the inclusion of non-ferrous impurity elements such as copper, zinc and tin, which are generally called Trump elements. Zinc and tin are mainly present on the surface of the plated steel sheet, whereas copper is mainly present as copper wire in the motor core of automobiles and home appliances, so it is possible to identify and separate at the crushing scrap stage to prevent contamination. Most effective in.

Conventionally, a regenerator has been used for processing. Waste such as automobiles and home electric appliances is first cut by a shredder called a shredder into long-sized scrap pieces of about several tens of millimeters, which are then clothed by wind-screening. , Non-metal pieces such as plastics, powders are removed, non-metal pieces and metal pieces are separated by eddy-current sorting, and iron and non-ferrous metal pieces are separated by magnetic sorting. The finally separated iron scrap is handed over to the steelmaker as an iron raw material.

However, since the copper core and the iron core are mechanically intertwined in the motor core, which is the largest source of copper contamination, the copper and iron cannot be separated even by magnetic separation, so that they are eventually conveyed on the belt conveyor. During the process, the workers were separated by visual identification and manual selection.

In such a manual identification work, it is difficult to expand the scrap processing amount, it is economically difficult to invest a large amount of labor cost, it is difficult to secure uniform quality of scrap after copper removal, and it is difficult to improve the working environment. There are problems such as. As a means of automating this work, a method of performing scrap automatic identification by laser beam irradiation has also been proposed (Dr. H.
-P Sattler: VDI BERICHTE NR. 934, 1991: 'Scrap sor
ting with Lazer-an automatic process for mixed non
-ferrous metals from automobile shredders').

[0006]

However, since the above-described method uses an expensive pulse laser irradiator, it is difficult to reduce the apparatus cost, and the laser and the spectroscope must be used in a bad environment. Practical problems have been that maintenance requires cost and manpower. As a method for automatically identifying scrap containing copper, a patent application has already been filed for an automatic identification method based on hue angle (Japanese Patent Application No. 5-135119; method for identifying scrap containing copper from a group of iron scraps). ).

However, in the method using only the hue angle, the hue angle becomes unstable in the portion where the chroma value in the scrap is low, so that the accuracy of discrimination between copper and iron is lowered. In order to efficiently separate copper-containing scrap in the separation process, it is necessary to identify each scrap in the identification process and obtain the position information of the copper-containing scrap and transmit it to the separation process. It was necessary to obtain the information of.

The problem to be solved by the present invention is to improve the identification accuracy of the copper-containing scrap by using the hue angle in order to realize an identification method with low equipment cost and maintenance cost, and to determine the position of the copper-containing scrap. To build an identification method so that the information can be transmitted to the separation process.

[0009]

The gist of the present invention is as follows. 1) An iron scrap group is imaged by a color television camera; 2) For each pixel in the image, a luminance value I, a saturation value S, and a hue angle value H are obtained from the RGB signal value of the point; 3) It is determined whether or not the saturation value of each pixel is greater than or equal to a preset threshold value for the image; if the saturation value of the pixel is greater than or equal to the threshold value, the hue angle of the pixel is determined. When the value is within the preset copper hue angle value range, it is determined that the pixel is copper; and when the hue angle value of the pixel is not within the preset copper hue angle value range, Determine that the pixel is not copper;
If the saturation value of the pixel is not greater than or equal to the threshold value, it is determined that the pixel is not copper; thereby obtaining a pixel in the image that is determined to be copper; 4) Using the luminance value I, the entire image Is subjected to a labeling process to identify individual scraps, and the total area St of the individual scraps, that is, the number of pixels occupied by the scraps is determined; the copper area S in the image area occupied by the scraps is calculated.
Obtain the total number of pixels determined to be Cu, that is, copper; 5) Ratio of copper area SCu to St for each scrap R =
SCu / St is calculated, and when the ratio R is equal to or more than a preset threshold value Rmin, the scrap is identified as a copper-containing scrap, and its center of gravity (X, Y), that is, the pixel of the center of gravity of the total area St is identified. Ask for a street address.

[0010]

As shown in FIG. 1, the iron scrap group 1 is connected to the light source 2
The area of the iron scrap group 1 where copper is required to be identified is imaged by the color television camera 3 in the state of being illuminated with. However, the light source is not always necessary if the image can be taken by the color television camera without the light source.

The RGB signal 4 of the color television camera is HS
By the I conversion device 5, the hue angle H (6), the saturation S (7),
And a luminance I (8) signal and transmitted to the identification processing device 9. The identification processing device 9 performs the image processing of, which will be described later, and outputs the barycentric position (X, Y) signal 10 of the scrap identified as containing copper.

The control device of the separating device (not shown) installed at the rear of the belt conveyor in the conveying direction receives the barycentric position (X, Y) signal 10 of the copper-containing scrap, and
A so-called tracking process is performed to correct a delay in the transportation time of the scrap, and an appropriate separation process is executed at the time when the scrap reaches the separation device.

Since the processing in the identification processing device 9 includes complicated image processing, real-time processing may not be possible in some cases. In that case, as shown in FIG. 1, the image pickup time signal 11 of the image is separated by the separation device. If it is transmitted to the control device, it is possible to perform an appropriate tracking process. Identification processing device 9
If real-time processing is possible with, it is possible to perform appropriate tracking processing only by correcting the transport time delay without transmitting the imaging time signal 11 of the image to the control device of the separation device. Needless to say.

The identification processing device 9 applies to the photographed image,
Copper-containing scrap is identified by performing the following processing. It is checked whether or not the saturation value of each pixel is equal to or larger than a preset threshold value; if the saturation value is equal to or larger than the threshold value, a hue angle value of the pixel is obtained, and the value is previously set. By comparing and referring to the correspondence relationship of the set hue angle value range of copper, if the hue angle is within that range, it is determined that the pixel is copper, and if it is not, it is not copper. When the saturation value is less than the threshold value, it is determined that it is not copper.

Labeling processing is performed on the entire image using the brightness value I to identify individual scraps.
The total area St of each scrap, that is, the number of pixels occupied by the scrap is obtained; the copper area SCu in the image area occupied by the scrap, that is, the total number of pixels determined to be copper by the processing; Ratio of SCu and St R = S
Calculating Cu / St, the ratio R is a preset threshold value R
When it is more than min, the scrap is identified as a copper-containing scrap, and its center of gravity (X, Y), that is, the total area S
Find the pixel address of the center of gravity of t.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 2 shows an example of the configuration of an identification device that implements the method of the present invention. In this example, in order to reduce the influence of light such as a fluorescent lamp in the factory, a dark room 22 is provided and the measurement is carried out in the dark room 22. A four-point light source 24 that illuminates scrap conveyed on a belt conveyor from four directions was used as a light source, and a CCD camera 23 was used as a color television camera. 4 point light source 2
4 has the effect of reducing the shadows produced on the scrap and its installation table, as compared with a light source that illuminates from one direction. The four-point light source is not necessarily required as long as the generation of shadows can be limited to a range that does not interfere with the identification process.

The image capturing device 27 is a CCD camera 23.
The R, G, and B signals 26 are input, converted into HSI signals 28, and output to the identification processing device 29. The identification processing device 29
Stores the copper hue angle range and saturation threshold,
It is determined whether or not all scraps in the image subjected to the labeling processing by the luminance I signal are copper-containing scraps, and the position information (X, Y) signal of the scraps identified as copper-containing is output. Further, in this embodiment, CC
In order to simplify the timing processing when the iron scrap image captured by the D camera 23 is captured in the identification processing device 29, the RGB signal 26 is temporarily stored in the image capturing device 27, converted into an HSI signal and output. .

Using the discriminating apparatus shown in FIG. 2, an automatic discriminating experiment was carried out for iron scrap contained in automobile shredder scrap and motor core scrap made of iron and copper. The shredder samples used in this experiment were to cut wastes such as automobiles and home electric appliances into shredders with a long size of about 80 mm, and by wind sorting, cloth, non-metallic pieces such as plastic, and powder. The non-metal pieces were separated from the non-ferrous metal pieces by eddy-current sorting, and the iron and non-ferrous metal pieces were further separated by magnetic sorting. Is a mixture of iron scrap consisting only of iron and motor core scrap containing copper. That is, the scrap samples used in this experiment have been identified and separated only by the visual inspection of a conventional worker and manual work.

Prior to the actual discrimination experiment, the following three types of preliminary experiments were carried out to determine the set values for this experiment. First of all, a preliminary experiment was carried out to determine the boundary value between the saturation value and the hue angle value for identifying copper. The same optical conditions as those used for identification (light source 24, operating conditions of CCD camera 23,
Under the setting conditions such as the darkroom 22, the distance between devices and the distance between target scraps, the distribution of hue angle values of the scrap copper portion, the scrap iron portion, and the base portion on which scraps are placed was measured.

Further, under the same conditions, the distribution of chroma values of the scrap copper portion, scrap iron portion, and belt conveyor portion was measured. From the results of these preliminary experiments, an appropriate range of saturation value S for judging copper and a range of hue angle value H of the copper part were found. These are
As shown in FIG. 4, the saturation values are (0.3 ≦ S ≦ 1.
0) (30), and the hue angle values are (0 ≦ H ≦ 52 ° and 329 ° ≦ H <360 °) (31). The discrimination values described below used the boundary values of the saturation value and the hue angle value for these discriminations. * Secondly, a so-called labeling process experiment was performed in which scraps in the image were individually recognized using the luminance signal. The belt conveyor is black and the luminance signal value is extremely low. A certain luminance signal threshold value is set to binarize the image, and the labeling process is performed by using the pixel continuity above the threshold value. As a result, iron scraps, motor core scraps containing copper, etc. were separated from the belt conveyor section, and individual scrap pieces could be easily recognized individually.

Although the brightness signal threshold value may be set in advance, the threshold value is automatically set to a threshold value of, for example, 25% of the average brightness of the entire image so as to cope with the case where the illumination light intensity varies. Needless to say, various conventionally proposed labeling processes such as setting can be used.

Third, the copper areas SCu and St, which are necessary for identifying whether or not the scrap contains copper.
An experiment for obtaining a threshold value Rmin of the ratio R = SCu / St is performed. That is, the R values of all the samples were measured using 100 samples of motor core scraps containing copper and 100 samples of iron scraps. As a result, in this experiment, it was found that the motor core scrap and the iron scrap can be completely distinguished by setting Rmin = 0.1.

That is, only two samples erroneously identified the motor core sample as the iron sample, and three erroneously identified the iron sample as the motor core sample.
It was only a sample. However, it goes without saying that the set value of Rmin may change depending on the degree to which the copper-containing scrap can be mixed into the iron scrap in the actual identification and separation process. Further, the barycentric position (X, Y) of the scrap identified as the copper-containing scrap can be easily calculated by simply averaging the pixel positions of the scraps using the binarized image obtained by the labeling process.

Identification processing device 29 used in an actual identification experiment
FIG. 5 is a flowchart showing a specific procedure of the identification processing performed by the. Motor core scrap 100 in the identification experiment
Randomly mix 60 pieces of iron scrap and 2,400 pieces of iron scrap
The identification experiment was performed by sending out onto a bell conveyor operating at a speed of (m / min). Since the identification device 29 used has the capability of performing high-speed image processing in real time, it was possible to easily perform tracking simply by correcting the conveyance time delay to the separation device. When the copper-containing scrap and the scrap separated into the iron scrap were analyzed, respectively, 17 iron scraps that were misidentified and separated in the copper-containing scrap and the copper-containing scrap that was misidentified and separated in the iron scrap were found. The number was four, and it was evaluated that the discrimination and separation performance was practically sufficient.

FIG. 6 is an example showing the result of individual recognition of scraps on the belt conveyor using the luminance (I) signal of a certain sampled image. In this case, since the belt conveyor is black and has extremely low brightness, it is possible to easily recognize the five scrap pieces individually by determining the continuity of the portion equal to or more than the preset threshold value. The white areas are recognized as individual scraps, and the total area St thereof is obtained.

FIG. 7 shows in white the pixels determined to be copper from the hue angle (H) signal and the saturation (S) signal of the image. The area SCu of the area determined to be copper is obtained for each scrap piece. As a result, the ratio of the area of the area determined as copper to the total area of each scrap piece R = SCu
/ St was calculated and the scrap pieces labeled 2 and 5 in the figure with R-values above the threshold Rmin = 0.1 for identifying copper-containing scrap were identified as copper-containing scrap.

From the corresponding color original images, it was found that the materials of each scrap piece were confirmed, 2 and 5 were crushed motor cores, and the rest were iron scraps composed of only iron. As shown in Fig. 7, even in the iron scraps 1, 3, and 4 there are some areas that are identified as copper, but this is due to the scrap surface coated with a red paint close to the color tone of copper and red rust. It was considered that the surface was attached or that the hue angle and saturation were within the range of copper due to the relationship between the illumination and the reflection angle. However, even if there is such a minute misidentification portion, it is possible to easily prevent the misidentification of the iron scrap and the copper-containing scrap by appropriately setting the threshold value Rmin of the ratio R.

[0028]

By utilizing the method for identifying copper in a scrap group using the method of the present invention, it is possible to automate and improve the accuracy of the identification work that has been conventionally performed by the visual inspection of workers, and as a result, the scrap processing amount can be increased. It is possible to solve the conventional problems such as easy expansion, no need to invest a large amount of labor cost, it is easy to secure uniform quality of scrap after copper removal, and it is easy to improve the working environment.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a device configuration that realizes an identification method based on the present invention.

FIG. 2 is an explanatory diagram of a device configuration used in a discrimination experiment.

FIG. 3 is a distribution diagram of a saturation value range of copper in the example.

FIG. 4 is a distribution chart of a hue angle value range of copper in the example.

FIG. 5 is a flowchart of an identification process used in the embodiment.

FIG. 6 is an identification diagram of an example of a labeling result by a luminance (I) signal.

FIG. 7 is an identification diagram of an example of a determination result of a copper portion based on a hue angle (H) signal and a saturation (S) signal.

[Explanation of symbols]

 1 Iron Scrap Group 2 Light Source 3 Color TV Camera 4 RGB Signal 5 HSI Converter 6 Hue Angle H Signal 7 Saturation S Signal 8 Luminance I Signal 9 Discrimination Processing Device 10 Center of Gravity (X, Y) Signal of Copper-containing Scrap 11 Imaging Time signal 22 Dark room 23 CCD camera 24 Four-point light source 25 Monitor TV 26 RGB signal 27 Image capture device 28 HSI signal 29 Discrimination processor 30 Saturation proper range for discrimination 31 Copper hue angle range

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shuji Naito 20-1 Shintomi, Futtsu-shi Nippon Steel Corporation Technical Development Division (72) Inventor Masahiro Ito 20-1 Shintomi, Futtsu Nippon Steel Co., Ltd. Technology Development Division

Claims (1)

[Claims] 1. A process for identifying and separating copper-containing scrap from a crushed iron scrap group conveyed on a belt conveyor. 1) An image of the iron scrap group is picked up by a color television camera; 2) Each pixel in the image. , The luminance value I, the saturation value S, and the hue angle value H are obtained from the RGB signal value of the point; 3) the saturation value of each pixel in the image is greater than or equal to a preset threshold value. If the saturation value of the pixel is greater than or equal to the threshold value, and the hue angle value of the pixel is within the preset copper hue angle value range, the pixel is determined to be copper. When the hue angle value of the pixel is not within the preset copper hue angle value range, it is determined that the pixel is not copper;
If the saturation value of the pixel is not greater than or equal to the threshold value, it is determined that the pixel is not copper; thereby obtaining a pixel in the image that is determined to be copper; 4) Using the luminance value I, the entire image Is subjected to a labeling process to identify individual scraps, and the total area St of the individual scraps, that is, the number of pixels occupied by the scraps is determined; the copper area S in the image area occupied by the scraps is calculated.
Obtain the total number of pixels determined to be Cu, that is, copper; 5) Ratio of copper area SCu to St for each scrap R =
SCu / St is calculated, and when the ratio R is equal to or more than a preset threshold value Rmin, the scrap is identified as a copper-containing scrap, and its center of gravity (X, Y), that is, the pixel of the center of gravity of the total area St is identified. Obtaining a street address; thereby identifying copper-containing scrap.