JP5011971B2 - Granular optical sorting method - Google Patents

Description

  The present invention relates to an optical sorting device that optically discriminates and sorts defective products contained in raw material granular materials such as cereals, millet grains, beans, or resin pellets, and particularly relates to determination means for determining defective products. Is.

  Conventionally, this type of optical sorting apparatus is configured to transfer and drop the raw material granular material in a strip shape and irradiate the raw material granular material with light, and RGB full-color detection light (RGB light) obtained from the raw material granular material. It is known that defective products are discriminated based on the above (hereinafter referred to as “full color discrimination of defective products”) and defective products are selected. In the optical sorting device, the determination unit includes a pass / fail determination reference table (hereinafter referred to as a “determination table”), and the determination of whether or not the detected light is a defective product refers to the determination table. It is performed instantaneously (so-called RGB determination method). In the determination table, determination results (non-defective product or defective product) are determined in advance for each of a plurality of detection lights (RGB light) that are assumed to be detected, in comparison with a set threshold value. Regarding the creation of the determination table, R (red) is, for example, 255 gradations, G (green) is 255 gradations, and B (blue) is 255 gradations for the light intensity (light quantity) of each color of the detection light (RGB light). , Each is divided into a plurality of categories, each R (G, B) category (combination) is compared with the set threshold value, and the judgment result (non-defective product or defective product) is a list. Yes (see FIG. 11).

  On the other hand, the method for determining the full color of the defective product includes the detected RGB light in addition to the method for determining good / bad from the RGB light simply detected from the raw material granular material as described above (so-called RGB light determination method). Is further converted to a value in a uniform color space (L * a * b * color system or L * u * v * color system) that is close to human vision (color sensation). There is a method (so-called uniform color space determination method) that performs the above (for example, Patent Document 1). Even in the case of this uniform color space determination method, a determination table as described above is configured in the determination unit, and the determination of good / bad detection light (RGB light) is performed instantaneously by referring to the determination table. It is. Regarding the creation of the determination table in the uniform color space determination method, for the light intensity (light quantity) of each color of the detection light (RGB light), for example, R (red) is 255 gradations and G (green) is 255 gradations, as described above. , B (blue) is divided into a plurality of divisions such as 255 gradations, and thereafter, a uniform color space (L * a * b * color system for each of the R, G, and B divisions (combinations) Etc.), each converted value is compared with a set threshold value, and the determination result (non-defective product or defective product) is listed (see FIG. 11). During the sorting operation, the light intensity combination of the colors of RGB light detected from the raw material granular material is instantaneously judged as good or bad with reference to the judgment table (for example, Patent Document 1).

JP 2005-186053 A

By the way, the optical sorting apparatus has the following problems. In other words, during the sorting operation, the threshold may be changed to change the sorting sensitivity. In this case, it is necessary to update the determination table (good / bad determination result) to the one corresponding to the changed threshold. Therefore, at least the recalculation time (several minutes) related to the update has to stop the sorting operation, which causes a problem that the efficiency of the sorting operation is lowered.
Accordingly, the present invention is directed to a technical problem that in view of the above problems, provides an optical sorting how that can change the threshold without stopping the sorting operation.

In order to solve the above problem, according to claim 1,
A transfer step of transferring the raw material granular material in a single layer;
An optical detection step of receiving RGB light from the raw material granular material by irradiating the raw material granular material discharged / dropped from the transfer step, and
The pass / fail judgment based on the RGB light detected in the optical detection step is based on a judgment table having pass / fail judgment information that is compared with a preset threshold value for each of a plurality of RGB lights assumed to be detected in the optical detection step. Determination process to be performed,
A sorting step for sorting the defective granular materials judged by the judging step;
In an optical sorting method having
The determination step includes two determination tables, and when the threshold value is changed during the sorting operation using one determination table, the determination step is based on the changed threshold value. The pass / fail judgment information for an arbitrary low-resolution gradation is recalculated and updated among all combinations of the other judgment tables not used for pass / fail judgment, and the sorting operation is stopped when the update is completed. Without switching the determination table used for pass / fail determination from one to the other, and then recalculating and updating pass / fail determination information for each of the combinations of the one determination table, when the update is completed Thus , the technical means of switching the determination table used for the pass / fail determination again from the other one to the one without stopping the sorting operation was taken.

Delete

According to claim 2 ,
The determination step includes a conversion step of converting the RGB light detected in the optical detection step into a value in a uniform color space of the L * a * b * color system or L * u * v * color system, The determination table has technical means that the determination table includes pass / fail determination information determined by comparing the threshold value set in the uniform color space with the converted values.

Delete

According to claim 1 of the present invention, two determination tables to be referred to when performing pass / fail determination based on the RGB light detected from the raw material granular materials are provided, and during normal sorting operation without changing the threshold value. One of the determination tables is used to make a pass / fail judgment. On the other hand, when the threshold value is changed during the sorting operation, the one that is not used for the pass / fail judgment based on the changed threshold value. The temporary determination table is created by recalculating (updating) only a portion of the gradations of an arbitrary low resolution among all combinations within a short time, and thereafter, the temporary determination table is created. Switch to continue the pass / fail judgment (selection operation), and recalculate (update) all combinations for the other judgment table that is not used for pass / fail judgment, and then create a formal judgment table. Switch to the judgment table Since continued determination, the reduction of the sorting operation efficiency becomes possible to change the threshold without stopping the sorting operation can be prevented. Thereby, the early creation / switching of the provisional determination table speeds up the reflection of the changed threshold value in the quality determination (selection). Further, when the formal determination table is completed, it is switched to this to make a pass / fail judgment, so that the pass / fail judgment can be made with the same high resolution as before the threshold change, and the pass / fail judgment can be made with a subtle difference in color.

Delete

Further, according to claim 2,
The RGB light detected by the optical detection means is provided with conversion means for converting the light in the uniform color space of the L * a * b * color system or L * u * v * color system, and the determination table includes: By configuring each conversion value and each pass / fail determination information determined by comparing the threshold value set in the uniform color space, the effect of claim 1 can be obtained also in the uniform color space.

Delete

Example 1:
Example 1 is an example relating to the so-called RGB light determination type optical sorting apparatus that performs determination by determining defective particles (defective products) based on RGB light detected from raw material granular materials.

  The overall configuration of the optical sorting apparatus 1 according to the present invention will be described (see FIG. 1).

Transportation means:
The optical sorting device 1 constitutes the raw material grain transfer means 2. In this embodiment, the belt type conveyor 2 is used as the transfer means 2. The belt-type conveyor 2 is configured so that an endless belt is stretched between left and right rotating shafts so as to be substantially horizontal, and the output of the motor 2a is transmitted to one of the rotating shafts so that the endless belt rotates in one direction. It is designed to rotate (arrow Y). A raw material supply unit 3 is disposed on one end side in the upstream direction of the belt type conveyor. The raw material supply unit 3 includes a raw material tank 4 and a vibration supply unit 5 disposed below the discharge port of the raw material tank 4. The vibration supply means 5 includes a vibration transfer rod 5a that feeds the raw material granular material to the endless belt in a single layer while vibrating, and a vibration portion 5b that supports the bottom surface of the vibration transfer rod 5a and applies vibration. Become. In addition, about the said transfer means, the thing of the chute | shoot type | mold which consists of a sloping flow-down hook may be sufficient.

Optical detection means:
An optical detection means 6 is disposed in the vicinity of the conveyance end portion of the belt type conveyor 2. The optical detection means 6 is composed of optical parts 6a and 6b, and the optical parts 6a and 6b are respectively provided on the upper side and the lower side across the fall trajectory G of the raw material granular material transferred and dropped from the belt type conveyor 2. It arrange | positions so that it may mutually oppose in a position. The optical units 6a and 6b respectively include a light source 7 for irradiating light to the optical detection position P on the fall locus G, a CCD camera 8 incorporating a CCD (charge coupled device) sensor as a light receiving element, and a background. (Background plate) 8a.

  Each of the CCD cameras 8 is adjusted and arranged so that the optical axis of the CCD camera 8 passes through the optical detection position P and comes into contact with the other background (background plate) 8a. The light source 7 is a white fluorescent lamp. The CCD sensor is a full color line CCD sensor (2048 elements, Si sensor) capable of receiving RGB (red, green and blue) light, and scans a horizontal line of the optical detection position P. The CCD camera 8 includes an amplifier (not shown) that amplifies the light (voltage analog signal) detected by the CCD sensor and an A / D converter (not shown) that converts the analog signal into a digital signal. Built in.

Sorting means:
At a position downstream of the optical detection position P and along the falling locus G, a nozzle 9 for blowing high-pressure air as a sorting unit is disposed. A plurality of the nozzles are arranged side by side. Each nozzle is connected to an electromagnetic valve via a high-pressure air conveying pipe, and each electromagnetic valve is connected to a high-pressure air supply source (not shown). Each electromagnetic valve is driven by an output signal from an injection drive circuit 19 described later.

Determination means: (see FIG. 2)
Next, a signal processing means (hereinafter referred to as “determination means”) 10 for processing an output signal from the CCD camera 8 will be described (see FIG. 2). The determination means 10 constitutes a preprocessing circuit 11 connected to the CCD camera 8, and a binarization processing circuit 12, a postprocessing circuit 13, and an arithmetic processing unit (CPU) 14 are connected via the preprocessing circuit 11. Connect and configure sequentially. The binarization processing circuit 12 is connected to a RAM (write / read storage unit) 15, and an arithmetic processing unit (CPU) 16 is connected to the RAM 15. The CPU 16 uses at least about 60 MHz operation. The pre-processing circuit 11 performs noise processing or the like on the output signal from the CCD camera 8 and uses a known one.

  A determination table (details will be described later) stored in advance is stored in the RAM 15, and the binarization processing circuit 12 refers to the determination table to output signals (detected RGB light) from the CCD camera 8. Is a circuit for performing pass / fail determination for. The binarization processing circuit 12 includes a storage unit that stores a truncation processing program to be described later.

  The post-processing circuit 13 is a circuit for specifying a defective pixel group in which a plurality of defective pixels determined by the binarization processing circuit 12 continuously form a lump (group). The arithmetic processing unit (CPU) 14 recognizes the defective pixel group determined by the post-processing circuit 13 as a defective particle and selects the corresponding nozzle 9 (electromagnetic valve) in order to select the defective particle by a high-pressure air jet. ). The arithmetic processing unit (CPU) 14 constitutes a delay circuit, and a selection signal is sent to the blast drive circuit 19 after a predetermined delay time.

  In addition, the arithmetic processing unit (CPU) 18 is connected to the binarization processing circuit 12 and the RAM 15, and the binarization processing circuit 12 has a determination table that is referred to in order to determine the quality of the RGB light. It is instructed whether to use the first determination table or the second determination table 17. An input unit 20 for setting and inputting a threshold value and the like is configured, and the CPU 18 is connected to the input unit 20.

Determination Table: See FIG. 11 A determination table (the first determination table or the second determination table 17) built in the RAM 15 will be described. In the determination table, for any of a plurality of detection lights (RGB light) that are assumed to be detected by the CCD camera 8, each is compared with a set threshold value in advance, and a determination result (good or defective) is obtained. It is determined. In the determination table, R (red) is set to, for example, 255 (8 bits (2 ⁸ = 256)) gradation (resolution), and G (green) is set to the 255th floor for the intensity (light quantity) of each color of the detection light (RGB light). G, B (blue) is divided into a plurality of divisions such as 255 gradations, and each R (G, B) division (combination) is compared with a set threshold value, and the pass / fail judgment result (pass / fail judgment) Information). One feature of the present invention is that two identical determination tables are built in the RAM 15. For convenience of explanation, one of the determination tables is a first determination table 16 and the other is a second determination table 17.

Operation of Example 1:
When driving (sorting operation) of the optical sorting device 1 is started, the raw material granular material flows down from the raw material tank 4 to the belt conveyor 2 through the vibration supply unit 5 and is layered by the belt conveyor 2 and transferred. And discharged from the end portion. The discharged raw material granular material falls in the form of a layer along the dropping locus G and is optically detected at the optical detection position P. The raw material granular material receives irradiation light from the light source (white fluorescent lamp) 7 at the optical detection position P, and RGB light from the raw material granular material is detected (scanned) by the CCD camera (full color line CCD sensor) 8. )

  The CCD camera (full color line CCD sensor) 8 amplifies the voltage analog detection signal of the received RGB light in units of pixels and converts it into a voltage digital signal, and the pixel data is sequentially transmitted through the preprocessing circuit 11. It is sent to the binarization processing circuit 12. The binarization processing circuit 12 makes a pass / fail judgment of each pixel data (RGB light) sent in sequence with reference to the first judgment table 16 (pass / fail judgment information) in response to a command from the CPU 18. Based on these pass / fail judgment results, the post-processing circuit 13 discriminates a defective pixel group and recognizes a defective particle, and the CPU 14 sets a predetermined delay time to drive an electromagnetic valve corresponding to the defective particle. Then, a drive signal is output to the blast drive circuit 19 and high pressure air is blasted from the corresponding nozzle 9 to select the defective product. The above is the explanation of the normal operation (selection operation without changing the threshold value).

Characteristic action (Example 1): See FIGS. 2 and 3 STEP1:
When the threshold value is changed during the sorting operation, first, a new threshold value is set and inputted to the CPU 18 from the input unit 20.
STEP2:
Next, the CPU 18 initializes all the pass / fail judgment information of the second judgment table 17 that is built in the RAM 15 and is not currently used for pass / fail judgment, and sets the changed threshold value and the luminous intensity combination of each section. Compare and recalculate to update. For the pass / fail judgment (selection operation) during the recalculation, the first judgment table 16 based on the threshold value before the change is continuously used.

STEP3:
When all the recalculations of the second determination table 17 are completed, the CPU 18 instantaneously switches the determination table to be referred to from the first determination table 16 to the second determination table 17 for the binarization processing circuit. Then, the pass / fail judgment (selection operation) is performed by the judgment table based on the changed threshold value. In this way, when changing the threshold value, the other determination table that is not used for the pass / fail determination is recalculated and updated, and the switching is instantaneously performed after the end of the update. There is no effect on sorting accuracy. Therefore, the sorting efficiency does not decrease.

Example 2:
The second embodiment relates to the same optical sorting apparatus of the RGB light determination method as that of the first embodiment, and a recalculation (updating) method of the determination table performed in accordance with the change of the threshold without stopping the sorting operation. This is an embodiment performed by a method different from 1.

  Since the configuration of the second embodiment is the same as that of the first embodiment except for the recalculation (updating) method of the determination table as described above, the description thereof is omitted. In the following, a description will be given focusing on characteristic actions and the like different from those of the first embodiment.

Characteristic action, etc. (Example 2): See FIG. 4 STEP1:
When changing the threshold value during the sorting operation, first, as in the first embodiment, a new threshold value is set and input from the input unit 20 to the CPU 18. Thereafter, Example 2 is different from Example 1.
STEP2:
The CPU 18 initializes all the pass / fail judgment information for the second judgment table 17 which is built in the RAM 15 and is not currently used for pass / fail judgment, and the low level among all 255 gradations (all combinations). Only 64 resolutions (64 combinations) of the resolution are recalculated and updated based on the changed threshold value (new threshold value) (see FIG. 5). This low resolution 64 gradation update is an update of the number of gradations (number of combinations) of a quarter of all 255 gradations, and the processing speed of the CPU 18 is relatively high at 60 megahertz operation. Update ends instantly.

STEP3:
When the update of the second determination table 17 is completed, the CPU 18 issues a switching command to the binarization processing circuit 12, and uses the first determination table 16 (corresponding to the old threshold value) currently used for the pass / fail determination. The display is switched to the second determination table 17 (only 64 gradations are updated corresponding to the changed threshold value).

  Simultaneously with this switching command, the CPU 18 issues a command to the binarization processing circuit 12 and executes the truncation processing program. The truncation processing program is a process performed to apply each pixel data (RGB light) sequentially sent to the binarization processing circuit 12 to 64 gradations corresponding to the changed threshold value in the second determination table 17. It is. As a result, the sorting operation (good / bad determination) is performed for the second determination table 17 (64 gradations corresponding to the changed threshold value), which is a temporary determination table updated corresponding to the changed threshold value. Only update) is performed continuously based on the changed threshold value.

STEP4:
On the other hand, when the determination table used for the pass / fail determination is switched to the second determination table 17 (only 64 gradations are updated corresponding to the changed threshold value), the first not currently used for the pass / fail determination. The determination table 16 is updated by the CPU 18. The update of the first determination table 16 is performed by initializing the pass / fail determination information of all 255 gradations and comparing / recalculating each of all 255 gradations (all combinations) with the changed threshold value. (See FIG. 6). Thereby, the first determination table 16 becomes a formal determination table in which all 255 gradations (all combinations) correspond to the changed threshold values.

STEP5:
When the update of the first determination table 16 (formal) is completed, the CPU 18 issues a switching command to the binarization processing circuit 12, and the second determination table 17 (a temporary determination table) currently used for the pass / fail determination. (Only 64 gradations are updated corresponding to the changed threshold value) is switched to the first determination table 16 (formal) and the selection operation (good / bad determination) is continued. At the same time as the switching, an instruction to stop execution of the truncation processing program is also issued, and each pixel data (RGB light) sent sequentially is not subjected to truncation processing, and the first determination table 16 (formally, all 255 gradations). Pass / fail) and pass / fail is determined.

  As described above, according to the recalculation (update) method of the determination table of the second embodiment, when the threshold value is changed during the sorting operation, based on the changed threshold value, the current one of the two determination tables is changed. Rather than re-calculating the determination table for each of the 255 gradations based on the changed threshold value, the determination table that is not used for the quality determination during the sorting operation is updated to 64 gradations. Only the minute (low resolution) is updated, and the temporary judgment table is completed within a short time, and the selection operation is continued by switching to this, and then the other judgment table is set for all 255 gradations (high resolution). Is updated, and a formal determination table is completed and switched to this to continue the sorting operation. Therefore, when the threshold value is changed during the sorting operation, the changed threshold value is immediately reflected in the pass / fail judgment, and the sorting operation is not stopped.

Example 3:
Unlike the first and second embodiments, the third embodiment employs a recalculation (updating) method of the determination table according to the present invention in the optical sorting apparatus 1 that performs the good / bad determination using the uniform color space determination method. It is an implemented example.

  The RGB light determination method of Example 1 and Example 2 is to determine good / bad based on the RGB light detected from the raw material granular material by the CCD camera 8. On the other hand, the uniform color space determination method according to the third embodiment converts the RGB light into a uniform color space value and compares the threshold value with the converted value in the uniform color space to determine pass / fail. is there. The uniform color space is generally known as a color system (L * a * b * color system or L * u * v * color system) that is considered to be close to human vision (color sense). (As determined by the CIE (International Commission on Illumination)).

  In the uniform color space determination method, the RGB light detected from the raw material granular material by the CCD camera 8 is set to the value of the uniform color space (L * a * b * color system or L * u * v * color system). The conversion process is performed by the CPU 18 of the signal processing means (determination means) 10 as in the first and second embodiments (see FIG. 7). The CPU 18 incorporates a known arithmetic expression for converting each pixel data of the RGB light into a value of the L * a * b * color system (or L * u * v * color system). As shown in FIG. 7 and FIG. 8, each pixel data (voltage digital signal = RGB value) of the detected RGB light is sequentially sent to the CPU 18 from the pre-processing circuit 11 (noise processing or the like). Thereafter, the RGB values are converted into XYZ values, and then the XYZ values are converted into Lab values (L * a * b * color system values). The Lab value calculated by the CPU 18 is sequentially sent to the binarization processing circuit 12, and the binarization processing circuit 12 makes a pass / fail determination with reference to a determination table built in the RAM 15.

  On the other hand, the threshold value can be set in the CPU 18 from the input unit 20 in the same manner as in the first and second embodiments, and the CPU 18 sets all 255 values for one of the determination tables based on the set threshold value (spherical). Gradation (all combinations) pass / fail judgment information (pass / fail judgment result) is calculated, and a judgment table is set before the sorting operation.

Characteristic action (Example 3):
When changing the threshold value during the sorting operation, the threshold value is first changed from the input unit 20 to the CPU 18 in substantially the same manner as in the first and second embodiments (see FIG. 9). . Then, as in the first embodiment, the CPU 18 recalculates and updates the other determination table that is not used for pass / fail determination for all 255 gradations based on the changed threshold value, and is completed. After that, the binarization processing circuit 12 is instructed to switch to the other determination table to determine whether it is acceptable. Since the switching is performed instantaneously, it can be performed during the sorting operation.

  On the other hand, the CPU 18 can also implement the recalculation (update) method of the determination table of the second embodiment (see FIG. 10). That is, the other judgment table not used for the pass / fail judgment is completed as described above, and a provisional judgment table (only 64 gradations out of all 255 gradations) is completed and switched to this to perform a sorting operation ( (Success / failure determination) is continued, and thereafter, all the 255 gradations of the other determination table are recalculated to update the respective pass / fail determination information, and thereafter, the other determination table is updated to the formal determination table in which the update of all other gradations is completed. Thus, it is possible to reflect the changed threshold value in the pass / fail judgment quickly by continuing the sorting operation (pass / fail judgment).

1 shows an overall configuration of an optical sorting apparatus according to the present invention. 1 is a block diagram of signal processing means (determination means) in Embodiment 1. FIG. The flow of the recalculation (update) method of the determination table in Example 1 is shown. The flow of the recalculation (update) method of the determination table in Example 2 is shown. (1) is a reference diagram conceptually showing the number of combinations in an arbitrary low resolution gradation (64 gradations) in the present invention, and (2) is the arbitrary low resolution gradation (64). 2 is a conceptual diagram of a (gradation) determination table. (1) is a reference diagram conceptually showing the number of combinations in an arbitrary high-resolution gradation (255 gradations) in the present invention, and (2) is the arbitrary high-resolution gradation (255). 2 is a conceptual diagram of a (gradation) determination table. The block diagram of the signal processing means (determination means) in Example 3 is shown. The conceptual diagram which converts RGB light into a uniform color space (L * a * b * color system) is shown. The flow of the recalculation (update) method of the determination table in Example 3 is shown. The flow of the recalculation (update) method of the determination table in Example 3 is shown. It is an example of a determination table.

1 Optical sorter 2 Belt conveyor (transfer means)
2a Motor 3 Material supply unit 4 Material tank 5 Vibration supply unit 5a Vibration supply rod 5b Vibration unit 6a Optical detection unit 6b Optical detection unit 7 Light source 8 CCD camera (full color line CCD sensor)
8a Background (background board)
9 Nozzle (sorting means)
10 Signal processing means (determination means)
11 Pre-processing circuit 12 Binary processing circuit 13 Post-processing circuit 14 Arithmetic processing unit (CPU)
15 Write / read memory unit (RAM)
16 First determination table 17 Second determination table 18 Arithmetic processing unit (CPU)
19 Fountain drive circuit G Falling locus P Optical detection position

Claims (2)

A transfer step of transferring the raw material granular material in a single layer;
An optical detection step of receiving RGB light from the raw material granular material by irradiating the raw material granular material discharged / dropped from the transfer step, and
The pass / fail judgment based on the RGB light detected in the optical detection step is based on a judgment table having pass / fail judgment information that is compared with a preset threshold value for each of a plurality of RGB lights assumed to be detected in the optical detection step. Determination process to be performed,
A sorting step for sorting the defective granular materials judged by the judging step;
In an optical sorting method having
The determination step includes two determination tables, and when the threshold value is changed during the sorting operation using one determination table, the determination step is based on the changed threshold value. The pass / fail judgment information for an arbitrary low-resolution gradation is recalculated and updated among all combinations of the other judgment tables not used for pass / fail judgment, and the sorting operation is stopped when the update is completed. Without switching the determination table used for pass / fail determination from one to the other, and then recalculating and updating pass / fail determination information for each of the combinations of the one determination table, when the update is completed A method for optically sorting granular materials, wherein the determination table used for pass / fail determination is switched again from one to the other without stopping the sorting operation . The determination step includes a conversion step of converting the RGB light detected in the optical detection step into a value in a uniform color space of the L * a * b * color system or L * u * v * color system, 2. The granular optical according to claim 1, wherein the determination table includes pass / fail determination information determined by comparing a threshold set in the uniform color space with the respective conversion values. Formula selection method.

Images