JP2008224619A - Method for discriminating quality of grain and device for selecting the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To automatically and surely discriminate all entries for quality of grains. <P>SOLUTION: In order to clearly discriminate a "split-skin" and a "navel" which have not been discriminated by the conventional concentration level judgment, the "navel" is specified based on the number of crests, the width between adjacent crests each other in addition to the concentration judgment. This can avoid to erroneously discriminate the "navel" as the "split-skin". Moreover, discrimination efficiency and accuracy can be improved. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a method for determining the quality of a bean grain that determines the quality related to the skin of the bean grain, and a bean grain sorting device that determines and determines the quality related to the skin of the bean grain.

  As a technology related to grain quality determination, a technology has been proposed that performs absolute evaluation that is not affected by differences in production area, variety, etc. by directly displaying raw data and function operation data measured by scanning as feature data. (See Patent Document 1).

In this patent document 1, an inspector visually confirms the display state on the LCD monitor, so that the quality of each determination item is determined. The feature data is raw data and function operation data that are uniquely determined from the photometric data, and the text message is only the result of comparison with the national average. There is no difference that can occur when pass / fail is determined. Pass / fail judgment depends on the inspector, but the inspector who is familiar with the difference in grain origin and variety will visually judge the correct data and text message for comparison with the national average. As a result, it is possible to determine the quality with high accuracy.
JP 2004-361333 A

  However, since the above prior art is intended for, for example, pass / fail discrimination in units of lots, it is advantageous in that the feature amount data is referred to improve the accuracy of discrimination by an expert visually. It is inefficient to determine whether each piece is good or bad.

  Accordingly, it is desirable to determine pass / fail by comparing the density value based on photometric data (read image data from a CCD sensor or the like) with a predetermined threshold value. However, the pass / fail determination value approximates the pass / fail density value. In some cases, it was not possible to make an accurate pass / fail judgment only by simple density value comparison.

  For example, black soybeans have a nearly black epidermis, and some of them have “navels” (“navels” are nutrient intakes for growing beans (nutrient intakes for the growth (ripening) of legumes). ))) However, when a defect (scratch defect) occurs that the epidermis is torn, the density of the defective skin part and the “navel” part is almost the same. It has become difficult.

  In view of the above facts, the present invention has an object to obtain a bean grain quality determination method and a bean grain sorting device capable of automatically and reliably determining the quality of all items of bean grain quality determination. .

  The invention according to claim 1 selects a special bean density part having a density value different from that of the normal bean grain density part based on the density difference of the image data on the surface of the bean grain, and the continuous density value of the special bean grain density part. Based on the nature, it is distinguished from the “navel” that belongs to the same concentration range, but takes in nutrients for growing bean grains, and distinguishes the presence or absence of a ruptured skin where there is a “crack” in which the epidermis has ruptured It is said.

  According to invention of Claim 1, a special bean grain concentration site | part is normally selected with respect to a bean grain concentration site | part. Based on the continuity of the density values of the special bean grain concentration part, it is distinguished from “navels” belonging to the same concentration range, and the presence or absence of a crack in which a “crack” exists is determined. That is, in claim 1, the distinction between “navel” and “crack” was found by the continuity of the special bean grain concentration site.

  The invention according to claim 2 is the invention according to claim 1, wherein a distance (interval) in which a predetermined concentration value continues from a change amount of the concentration value of the special bean grain concentration site is calculated, and the distance (interval) is calculated. ) Is continuous for a predetermined period or more, it is determined that the concentration is a special bean grain concentration due to the “crack”, and there are intermittently a plurality of special bean density concentration sites where the continuous distance (interval) is less than a predetermined value, And when the ratio of the distance (interval) in the density | concentration of the adjacent special bean grain density | concentration site | parts exists in the predetermined range, it is characterized by determining with a "navel."

  According to the second aspect of the present invention, there is shown an embodiment of discrimination based on continuity, and when the distance (interval) of the concentration value of the special bean grain concentration site is continuously longer than a predetermined value, It is discriminate | determined that it is the special bean grain density | concentration, and the distance (interval) in the density | concentration of the special bean grain density | concentration parts adjacent to each other where there exist intermittently a plurality of special bean density density | concentrations whose said continuous distance (interval) is less than predetermined When the ratio is within a predetermined range, it can be determined as “navel”.

  The predetermined range is within a predetermined range of 1 or larger and the inverse thereof. In addition, when comparing the ratio between the bulges (widths) of the special bean grain concentration sites themselves, the continuous distance (interval) is compared with the ratio of the distance (length between the highest concentration points) of the adjacent special bean concentration sites. In this case, the case of comparing the ratio of the intervals between the adjacent special bean grain concentration sites (intervals of predetermined concentration value levels) is included.

  According to a third aspect of the present invention, in the first or second aspect of the present invention, the outline of the bean grain is detected based on the density difference of the image data on the surface of the bean grain, and the maximum vertical dimension and the maximum in the outline are detected. It is characterized by determining whether or not there is a flat defect based on the ratio to the horizontal dimension, and determining whether or not there is a defective endothelial adhesion based on the density difference of the image data on the surface of the beans.

  According to the invention of claim 3, by using the density of the image data, the outline of the bean grain is detected based on the density difference of the image data on the surface of the bean grain, and the maximum vertical dimension and the maximum horizontal dimension in the outline are It is possible to determine whether or not there is a flat defect based on the ratio, and to determine whether or not there is a poor endothelial adhesion based on the density difference of the image data on the surface of the beans.

  The invention according to claim 4 is for determining whether or not the skin is particularly good with respect to the bean grains encased in the epidermis of a predetermined concentration range, where the “navel” for incorporating nutrients for growth is exposed on the surface. Then, a bean grain sorting device for sorting, an image pickup unit that scans and images the surface of the bean grains while performing relative movement in a direction orthogonal to the scanning direction, and an image picked up by the image pickup means And classifying means for classifying the image data into a normal epidermal density part and a special epidermal density part having a predetermined density difference with respect to the normal epidermal density part, and the special display density classified by the classification means Among the parts, the ratio of the distance (interval) between the parts that are intermittent in the scanning direction and adjacent to each other is determined by the determination means and whether the determination means affirmed, Includes special epidermal concentration Identifying means for identifying the presence of the “navel” in the bean grain and identifying the presence of the “crack” in which the epidermis has ruptured in the bean grain that includes the special epidermis concentration site. And a sorting unit that sorts out defective items of other items including at least beans with a “crack” identified by the identifying unit, and extracted based on an analysis result of image data captured by the imaging unit. Have.

  According to the fourth aspect of the invention, for example, scanning for imaging is repeated while moving the beans, and image data obtained by imaging by scanning is analyzed.

  By this analysis, the normal skin concentration part and the special skin concentration part having a predetermined density difference with respect to the normal skin concentration part are classified.

  It is determined whether or not the ratio of the distances (intervals) between adjacent parts that are intermittent in the scanning direction among the classified special display density parts belongs to a predetermined range.

  If the determination is affirmative, it is specified that the “navel” is present in the beans containing the special epidermis concentration site.

  Further, when the determination is negative, it is specified that a “crack” in which the epidermis has ruptured exists in the beans including the special epidermis concentration site.

  The sorting means sorts other item defective products that contain beans with “fissure” and that are extracted based on the analysis result of the image data captured by the imaging means.

  Other items include overall color density, endothelial adhesion, flatness, and the like.

  As a result, it is possible to automatically and accurately determine pass / fail in units of one bean and to automatically sort.

  As described above, the present invention has an excellent effect that the quality can be automatically and surely determined for all items of the quality determination of the beans.

(Device configuration)
1 and 2 show a schematic configuration of a bean grain sorting apparatus 10 according to the present embodiment.

  The bean grain sorting apparatus 10 is configured around a cylindrical rotary valve 12.

  The rotary valve 12 is rotated at a constant speed in the counterclockwise direction of FIG. 2 by the driving force of the drive motor 116 (see FIG. 8).

  As shown in FIG. 3, streak-like grooves 14 parallel to the axial direction are provided on the circumferential surface of the rotary valve 12 almost uniformly over the entire circumference. A partition wall 16 is formed inside the groove 14 so as to partition the axial direction, and each region partitioned by the partition wall 16 serves as a bean container 20 that stores and holds the bean grains 18.

  In addition, a marker 22 is attached to one end side in the axial direction of the peripheral surface of the rotary valve 12 according to the presence or absence of the groove 14. In the present embodiment, the marker 22 is attached to a portion where the groove 14 does not exist in the axial direction.

  As shown in FIG. 1, a vibration feeder unit 24, an imaging unit 26, and a sorting and discharging unit 28 are provided around the rotary valve 12 in order from the upstream side along the flow of the beans 18. The sorting and discharging unit 28 is divided into a defective target opening 28A and a non-defective target opening 28B (see FIG. 2).

  The vibration feeder unit 24 is arranged on the rotary valve 12 in the right direction (direction from 2 o'clock to 3 o'clock) in FIG. The vibration feeder portion 24 is provided with a hopper portion 30 so that a large number of beans 18 are put in. The beans 18 put in the hopper 30 are vibrated by the vibration source 114 and are sequentially sent out to the alignment table 32.

  The alignment table 32 is formed with a plurality of guide grooves (not shown) (not shown) in the present embodiment, and each of the leading ends thereof has a bean accommodating portion 20 in the groove 14 along the axial direction of the rotary valve 12. It is arranged to correspond to. Thereby, the bean grains 18 are accommodated one by one in the bean accommodating portion 20 of each groove 14 as the rotary valve 12 rotates counterclockwise in FIG. 2.

  The imaging unit 26 is provided in the upward direction (12 o'clock direction) in FIG.

  In the housing of the imaging unit 26, there is a CCD line sensor 26A, an optical system (for example, an imaging lens unit (lens of a single-lens reflex camera) including a convex lens that compensates for aberration in the longitudinal direction of the CCD line sensor 26A), or for each pixel. 26B is housed.

  Here, the relative position relationship between the CCD line sensor 26A and the optical system 26B is determined so that the bean grain 18 accommodated in the bean accommodating portion 20 of the rotary valve 12 becomes a focal point during imaging by the CCD line sensor 26A. ing.

  The CCD line sensor 26A scans in the pixel array direction (that is, the axial direction of the rotary valve 12) while the rotary valve 12 is rotating in the counterclockwise direction of FIG. As a result, in the present embodiment, six (one row) beans 18 are read with a total of seven imaging scan lines, and image data (density data) is generated.

  Image data (density data) generated by reading with the CCD line sensor 26A is sent to the controller 34 for image analysis (details will be described later).

  An injection valve 38 </ b> A for the air nozzle 38 is provided in the inner space of the rotary valve 12.

  The injection valve 38 </ b> A is provided so as to face the through hole provided in the bottom of each bean container 20 of the groove 14 when the rotary valve 12 reaches a predetermined rotational position. That is, the six injection valves 38 </ b> A are arranged in a line along the axial direction of the rotary valve 12.

  The predetermined rotational position of the rotary valve 12 is a position where the target groove 14 is opposed to the defective target opening 28A. At this position, a rotational position detection sensor 40 that detects the rotational position of the rotary valve 12 is used. Is provided.

  The injection valve 38A is opened / closed by excitation / non-excitation of the injection solenoid 115 (see FIG. 8). A predetermined internal pressure is applied in advance to the piping of the air nozzle 38 by driving an air compressor 104 (see FIG. 8), and when the injection valve 38A is opened, air is discharged with a force according to the internal pressure.

  Here, when the target groove 14 of the rotary valve 12 reaches the predetermined rotation position, when the injection valve 38A is appropriately opened, it is stored in the bean storage portion 20 corresponding to the opened injection valve 38A. The air discharged from the bean grains 18 is blown off to the defective target opening 28A and accommodated as defective bean grains.

  That is, it has a structure in which it is possible to select whether or not to blow off defective products in units of individual bean containers (one bean grain).

  Note that the bean grains 18 accommodated in the bean accommodating portion 20 facing the injection valve 38A maintained in the closed state are rotated by the rotary valve 12 in the counterclockwise direction of FIG. It falls to the part 28B and is accommodated as non-defective beans.

  The controller 34 includes a drive motor 116 that rotates the rotary valve 12, a vibration source 114 of the feeder unit 24, an injection solenoid 115 that drives the injection valve 38A of the air nozzle 38, a marker detection sensor 118 that detects the marker 22, and a rotary A rotational position detection sensor 40 for detecting the rotational position of the valve 12 is connected and controls driving. An operation panel 42 is connected to the controller 34. The operation panel 42 includes at least an operation switch 42A for instructing an operation start and a stop switch 42B for instructing an operation stop. An input system such as a main power switch 42C and a monitor 42D for displaying an operation state of each unit. And.

(Characteristics specific to black soybeans, as well as good / bad judgment)
In the bean grain sorting apparatus 10 of the present embodiment, it is possible to perform appearance density inspection and sorting for various bean grains 18 within a predetermined outer dimension. Here, taking “black soybeans” 18 as an example of the beans 18, the selection of defective appearance and taste products of the black soybeans 18 will be described.

  First, as shown in FIG. 4, the black soybean 18 has a high concentration of epidermis close to black, and a non-defective product is called sized (see FIG. 4A). FIG. 4B is an example of a defective product, and shows black soybeans 18 with poor white skin adhesion in which the white skin inside the pods is attached to the epidermis. FIG. 4C is another example of a defective product, and shows black soybeans 18 with a cracked skin whose skin has been torn.

  (1) The black soybean 18 is entirely covered with the epidermis and has a density value equal to or higher than a predetermined value close to black (normal epidermis density), for example, an average density value of a planar image (scanned image), By comparing with a predetermined reference density value, the suitability of the color density of the black soybean 18 can be determined.

  (2) The black soybean 18 impairs the appearance quality due to the adhesion of the endothelium to the surface. The endothelium has an extremely lower concentration (close to white) than the normal epidermal concentration value (normal epidermal concentration).

  Therefore, it is possible to easily determine the presence or absence of endothelium adhesion by comparing the density value of the planar image of black soybean 18 with a relatively rough threshold setting (see FIG. 5).

  (3) The black soybean 18 has a flatness as a good / defective product selection element as a consistent appearance quality.

  Flatness means the ratio of the maximum vertical dimension and the maximum horizontal (width) dimension of black soybean 18 in the present embodiment. For example, the maximum value in the vertical direction and the maximum value in the horizontal (width) direction of the outline (boundary) are extracted from the image data obtained by the planar image (by scanning a plurality of times), the ratio is calculated, and the predetermined value is determined in advance. Black soybeans 18 corresponding to a range (a predetermined range greater than 1) are selected as non-defective products (see FIG. 6).

  (4) The surface of black soybean 18 is generally covered by the epidermis, and it is normal (normal) that the contents cannot be seen, but when the surface is torn and the contents are exposed (hereinafter, the surface is torn) Is referred to as “crack”), which impairs the appearance quality as well as the taste quality.

  By the way, the density value (special epidermal density) of the part where “crack” has occurred is lower (whiter) than the epidermal density (normal epidermal density). In the case of two types of rupture sites, it is possible to detect a defect due to “fissure” by performing the same process as the determination of the presence or absence of endothelium attachment (see FIG. 7).

  However, the black soybean 18 has a “navel” on its surface. “Navel” is an indispensable part for intake of nutrients for growing beans (incorporation of nutrients for growth (ripening) of legumes). This “navel” has a slightly lower concentration (closer to white) than the skin concentration (normal skin concentration). That is, assuming that the normal epidermis is the normal display concentration site, the “crack” and the “navel” can be said to be almost the same concentration site (special epidermal concentration site).

  For this reason, it is difficult to distinguish between “crack” and “navel” by simple density difference discrimination. In view of this, the controller 34 of the bean grain sorting apparatus 10 according to the present embodiment stores the bean grain pass / fail judgment algorithm capable of judging the quality of all the black soybeans 18 of (1) to (4) as an execution program.

(Control function)
FIG. 8 is a functional block functionally showing the control executed by the controller 34 according to the bean grain quality determination algorithm. Note that this functional block diagram does not limit the hardware configuration and is functionally classified.

  The main power switch 42C of the operation panel 42 is connected to the power management unit 100 of the controller 34. When the main power switch 42C is turned on, the power management unit 100 supplies power to the controller 34 and each unit drive system. To do.

  Further, the operation switch 42 </ b> A and the stop switch 42 </ b> B are connected to the activation control unit 102. The activation control unit 102 starts driving the air compressor 104 of the air nozzle 38 based on the operation of the operation switch 42A, and instructs the main control unit 106 of the controller 34 to execute control.

  In the air compressor 104, the pressure in the air nozzle 38 is controlled so as to be a predetermined internal pressure by original regulation control.

  The main control unit 106 collectively controls a series of operations such as feeding black soybeans 18, imaging, pass / fail judgment, and sorting.

  When the stop switch 42B is operated, the start control unit 102 instructs the air compressor 104 and the main control unit 106 to stop the current operation state.

  The main control unit 106 is connected to a monitor 42D of the operation panel 42, and sends operating state information to the monitor 42D.

  The driving state information is, for example, set initial setting information (threshold value, etc.), quantity information of black soybeans 18 currently input, classification ratio information, and the like.

  The main control unit 106 is connected to each of the vibration feeder drive control unit 108, the rotary valve drive control unit 110, and the imaging control unit 112 of the imaging unit 26.

  The vibration feeder drive control unit 108 is connected to the vibration source 114 of the vibration feeder unit 24. The vibration feeder drive control unit 108 drives the vibration source 114 based on a command from the main control unit 106 and arranges the black soybeans 18 put into the hopper 30 one by one in each of the six grooves, Move in the direction of the rotary valve 12.

  The rotary valve drive control unit 110 is connected to the drive motor 116 of the rotary valve 12. The rotary valve drive control unit 110 drives the drive motor 116 based on a command from the main control unit 106 to rotate the rotary valve 12 in the counterclockwise direction of FIG. 2 at a constant speed.

  In addition, a marker detection sensor 118 and a rotational position detection sensor 40 are connected to the rotary valve drive control unit 110. The marker detection sensor 118 detects the marker 22 affixed to the peripheral surface of the rotary valve 12, and the rotational position detection sensor 40 detects the position at which the selection target groove 14 faces the injection valve 38A. It is. For this reason, the marker sensor 118 is connected to an image analysis unit 120 described later, and the rotational position detection sensor 40 is connected to an air nozzle drive control unit 122 described later.

  The imaging control unit 112 receives a command from the main control unit 106 (imaging timing information based on the rotational position of the rotary valve 12), and executes imaging (scanning) by the CCD line sensor 26A.

  Scanning imaging information (image data) from the CCD line sensor 26A is sent to the image analysis unit 120 line by line. In the present embodiment, seven scans are performed on six (one row) black soybeans 18 in relation to the rotational speed of the rotary valve 12. Note that the number of scanning lines is not limited. In addition to the CCD line sensor 26A, the rotary valve 12 is stopped at every pitch of the groove 14 (may be driven at a low speed in relation to the shutter speed), and an image is picked up by a CCD area sensor in which images are arranged in the vertical and horizontal directions. You may do it.

  The image analysis unit 120 determines the quality of the black soybean 18 by comparing the density with a threshold or determining the continuity of the density change based on the captured image data, and the determination result ( The quality of each bean container 20 of the rotary valve 12 is sent to the air nozzle drive controller 122 (details will be described later).

  In the air nozzle drive control unit 122, based on the rotational position detection sensor 40, it is determined whether or not the groove 14 to be selected is opposed to the injection valve 38 </ b> A, and the bean container 20 unit partitioned by the partition wall 16. The opening and closing of the injection valve 38A is controlled.

  As a result, the black soybeans 18 determined to be defective exceed the non-defective target opening 28B by the air jetted from the air nozzle 38A, are accommodated in the defective target opening 28A, and are processed as defective products.

  The non-defective black soybeans 18 accommodated in the bean container 20 that was not opened by the injection valve 38A are accommodated in the non-defective target opening 28B by their own weight as the rotary valve 12 continues to rotate, and are processed as non-defective products. The

(Image analysis (quality determination) in the image analysis unit 120)
(1) Color density suitability determination The average density value of the planar image (scanned image) is compared with a predetermined reference density value, and if the ratio is less than or equal to the reference density value, the fourth operation flag for air nozzle operation Is turned on (see step 242 in FIG. 9).

(2) Judgment of presence / absence of endothelium adhesion The density value of the planar image is compared with a set threshold value. (See step 222).

(3) Flatness level determination It is determined whether or not the ratio of the maximum vertical dimension to the maximum horizontal (width) dimension belongs to a predetermined range (a predetermined range greater than 1). The third operation flag is turned ON (see step 238 in FIG. 9).

(4) Crack defect determination The average density value of the planar image (scanned image) is compared with a predetermined reference density value, and it is determined whether or not a region lower than the reference density value continues for a predetermined number of times. If the area below the density value continues for a predetermined period or more, the second operation flag for air nozzle operation is turned ON (see step 226 in FIG. 9).

  If normal epidermis is the normal display concentration site, the “crack” and “navel” are almost the same concentration site (special skin concentration site). It is determined whether or not.

  That is, the image data of the black soybean 18 is first-order differentiated using the same image data as that at the time of the crack defect determination, and a density change peak is obtained based on the result, and there are a plurality of density change peaks. When the ratio of the distances (intervals) between the peaks is within a predetermined range, it is determined that it is “navel” and the navel flag is turned on. Note that a threshold value may be provided for the distance between peak peaks, and a procedure may be added in which a “navel” is not considered when the number of pixels exceeds a certain value.

  Here, the relationship between the second operation flag and the navel flag is given priority over the navel flag. Therefore, even if the second operation flag is ON, if the navel flag is ON, the air nozzle operation is avoided.

  The operation of this embodiment will be described below with reference to the flowcharts of FIGS.

  FIG. 9 shows a process flow (algorithm) based on a control program stored in the controller 34.

  First, when the operation switch 42A is turned on in step 200, the process proceeds to step 202, where initial setting / confirmation for starting sorting is performed, and the process proceeds to step 206.

  In step 206, capture of an image read by the CCD line sensor 26A is started, and the process proceeds to step 208.

  In step 208, it is determined whether or not the stop switch 42B has been operated. If the determination is negative, the process proceeds to step 210, image data (waveform) is taken from the CCD line sensor 26A, and image analysis is executed. Move to 214. In step 208, if the stop switch 42B is operated, the sorting ends even during the sorting operation.

  In step 214, the presence or absence of the marker 22 is confirmed. If it is determined that there is no marker, the process proceeds to step 216, and the line scanning of one groove 14 in the rotary valve 12 (here, seven times scanning) is completed. It is determined whether or not. That is, the selection operation of the non-defective product and the defective product based on the determination result is executed for each scan for the presence / absence of endothelium and for the flatness and the color density for every scan (seven scans).

  If a negative determination is made in step 216, that is, if all scanning has not been completed, the routine proceeds to step 218. In step 218, “navel” discrimination processing control is executed. The “navel” discrimination process control in step 218 will be described later.

  When the “navel” discrimination control at step 218 is finished, the routine proceeds to step 220, where it is judged whether or not there is an image that is whiter than the threshold level for judging the endothelial adhesion (determination of endothelium adhesion). When an affirmative determination is made at step 220, the routine proceeds to step 222, where the first operation flag indicating whether or not to open the injection valve 38A of the air nozzle 38 is turned ON ("ON" means the injection valve 38A is opened). ) And the process proceeds to step 224. On the other hand, if a negative determination is made in step 220, the process proceeds to step 224.

  In step 224, it is determined whether or not a plurality of pixels corresponding to the special epidermis concentration site are continuous (determination of presence or absence of a fissure). If an affirmative determination is made in step 224, the routine proceeds to step 226, where a second operation flag indicating whether or not to open the injection valve 38A of the air nozzle 38 is turned ON ("ON" means that the injection valve 38A is opened). ) And the process proceeds to step 228. If a negative determination is made in step 224, the process proceeds to step 228.

  In step 228, the values having the maximum vertical dimension and the maximum horizontal (width) dimension in the line scan so far are updated and stored. (Including the skin concentration part) and the entire area of the black soybean 18 in the scanning range are accumulated and stored, and the process proceeds to Step 232.

  In step 216, the process also proceeds to step 232 when an affirmative determination is made, that is, when seven scans are completed.

  If the determination in step 214 is affirmative, that is, if the marker 22 is detected, the process proceeds to step 234.

  In step 234, it is determined whether or not the marker has just passed, that is, whether or not the marker 22 has been detected for the first time. If the determination is negative, the process proceeds to step 232.

  If an affirmative determination is made in step 234, the process proceeds to step 236, where the ratio of the maximum vertical dimension and maximum horizontal (width) dimension stored in step 228 exceeds a predetermined range (one or more predetermined ranges). If it is determined whether or not (bean grain flatness defect determination) and an affirmative determination is made, the routine proceeds to step 238, where a third operation flag indicating whether or not to open the injection valve 38A of the air nozzle 38 is turned ON ("ON"). Means opening of the injection valve 38A), and the process proceeds to step 240. If a negative determination is made in step 236, the process proceeds to step 240.

  In step 240, it is determined whether or not the ratio of the white area stored in step 230 to the black area obtained by removing the white area region from the total area of the black soybean 18 is equal to or greater than a predetermined threshold value (for the whole bean grain). If the determination of color density is good), if an affirmative determination is made, the routine proceeds to step 242 where the fourth operation flag indicating whether or not to open the injection valve 38A of the air nozzle 38 is turned ON ("ON" is the opening of the injection valve 38A). The process proceeds to step 244. On the other hand, if a negative determination is made in step 240, the process proceeds to step 244.

  In step 244, the maximum vertical dimension data, maximum horizontal (width) dimension data, white area data, and total area data (including black area data obtained by the difference between the total area data and the white area data) are cleared. Migrate to

  In step 232, the state detected by the rotational position detection sensor 40 is recognized, and it is determined whether or not the rotary valve 12 has reached a predetermined rotational position. If a negative determination is made, the process returns to step 208 and the above steps are repeated.

  If an affirmative determination is made in step 232, the process proceeds to step 246, in which one of the first operation flag is ON, the third operation flag is ON, and the fourth operation flag is ON. It is determined whether or not.

  If a negative determination is made in step 246, the process proceeds to step 248 to determine whether or not the second operation flag is ON. If a negative determination is made, the process returns to step 208 to repeat the above steps.

  If the determination at step 248 is affirmative, the routine proceeds to step 250 where it is determined whether or not the navel flag is OFF (no “navel”). If a negative determination is made in step 250, the process returns to step 208.

  If an affirmative determination is made in step 246 or step 250, it is determined that there is a defect, the process proceeds to step 252, the injection valve 38A of the air nozzle 38 is opened, and the black soybean 18 is blown off to the defective target opening 28A. Return to.

  Next, the “navel” determination control routine in step 218 of FIG. 9 will be described according to the flowchart of FIG.

  In step 260, the bean image data is extracted from the scanned image data of one line, and then the process proceeds to step 262 to execute the primary differential operation, and the process proceeds to step 264.

  In step 264, a steeply inclined region in which a change amount equal to or larger than a predetermined value is extracted in units of beans is extracted, and then in step 266, a mountain corresponding to the concentration is created. In this case, the maximum point side is the minimum density.

  In the next step 268, it is determined whether or not there are a plurality of created mountains. If a negative determination is made, it is determined that it is not “navel”, the process proceeds to step 270, and the “navel” flag is turned OFF. Ends.

  If the determination in step 268 is affirmative, the process proceeds to step 272, where it is determined whether or not the ratio of the distances (intervals) between adjacent mountains is within a predetermined range. If a negative determination is made in this step 272, it is determined that it is not “navel”, the process proceeds to step 270, the “navel” flag is turned OFF, and this routine ends.

  If an affirmative determination is made in step 272, it is determined that it is a "navel", the process proceeds to step 274, the navel flag is turned on, and this routine ends.

  In a specific example in accordance with the flowchart of FIG. 10, the image data corresponding to the width of the black soybean 18 obtained by scanning one line is set to about 250 pixels, and first-order differential calculation is performed on this pixel array. (Change amount of 5 pixel pitch).

  After that, a sample in which the calculated differential value (absolute value) continues for a predetermined number or more is extracted. At this time, a so-called right-up shoulder region and a right-down shoulder region are paired, resulting in the presence of one pair or multiple pairs.

  Here, when there are a plurality of pairs, it is determined that there are a plurality of mountains having density changes, and when the distance (interval) between the adjacent mountains is within a predetermined range, it is determined that it is a “navel”. . If this “navel” is determined, the navel flag is turned on.

  FIG. 11 is a reference diagram, and is a flowchart showing an implementation algorithm along the control flow shown in the flowchart of FIG.

  In FIG. 11, an algorithm for extracting a portion where each step indexed by AA has a large change amount is extracted, an algorithm for detecting a peak due to a density difference by each step indexed by BB, and a mountain adjacent to each step indexed by CC This is an algorithm for determining the presence or absence of a “navel” based on the ratio of the widths.

  As described above, in the present embodiment, in order to clearly discriminate between “crack” and “navel” that could not be discriminated by the conventional density value level judgment, in addition to the density value level judgment, the density difference The “navel” is specified based on the number of mountains according to the ratio and the ratio of the widths of adjacent mountains. For this reason, it is possible to avoid misidentifying it as “fissure” even though it is “navel”. The discrimination efficiency and the discrimination accuracy can be improved.

It is a schematic block diagram of the bean grain discrimination | determination apparatus which concerns on this Embodiment. It is an enlarged view of the principal part of the bean grain discrimination | determination apparatus which concerns on this Embodiment. It is a perspective view of a rotary valve. It is a front view of the black soybean as a bean grain, (A) is sized, (B) is poor endothelium adhesion, (C) shows poor tear. It is a chart which shows the discrimination algorithm of endothelial defect. It is a chart which shows the discrimination algorithm of flat (aspect ratio) defect. It is a chart figure which shows the algorithm which distinguishes "a cleft" and a "navel". It is a functional block diagram of the control performed according to the bean grain pass / fail judgment algorithm in the controller. It is a flowchart which shows the flow (algorithm) based on the control program memorize | stored in the controller. It is a control flowchart which shows the subroutine of step 218 of FIG. FIG. 11 is a mounting flowchart for mounting the processing of FIG. 10 on a controller. FIG.

Explanation of symbols

10 Bean Grain Sorting Device 12 Rotary Valve 14 Groove 16 Partition Wall 18 Bean Grain (Black Soybean)
20 Bean container 22 Marker 24 Vibrating feeder unit 26 Imaging unit 26A CCD line sensor (imaging means)
26B Optical system 28 Sorting and discharging section 28A Defective target opening 28B Non-defective target opening 30 Hopper section 32 Alignment table 34 Controller 38 Air nozzle 38A Injection valve 40 Rotation position detection sensor 42 Operation panel 42A Operation switch 42B Stop switch 42C Main power switch 42D Monitor 100 Power management unit 102 Start control unit 104 Air compressor 106 Main control unit 108 Vibration feeder drive control unit 110 Rotary valve drive control unit 112 Imaging control unit 114 Vibration source 115 Injection solenoid 116 Drive motor 118 Marker detection sensor 120 Image analysis unit ( Classification means, discrimination means, identification means)
122 Air nozzle drive control unit (sorting means)

Claims (4)

Based on the density difference of the image data on the surface of the bean grain, a special bean density part having a density value different from the normal bean grain density part is selected,
Based on the continuity of the concentration value of the special bean grain concentration site, a crack that belongs to the same concentration range but has a “crack” in which the epidermis ruptures is distinguished from a “navel” that takes in nutrients for growing bean grains. A method for determining whether a bean grain is good or bad, wherein the presence or absence of defective skin is determined. Calculate the distance (interval) that the predetermined concentration value continues from the amount of change in the concentration value of the special bean grain concentration site,
When the distance (interval) is continuous for a predetermined distance or more, it is determined that the concentration is a special bean grain concentration caused by the “crack”,
There are intermittently a plurality of special bean concentration sites where the continuous distance (interval) is less than a predetermined value, and the ratio of the distances (intervals) in the concentration between adjacent special bean concentration sites is within a predetermined range. In this case, the bean quality determination method according to claim 1, wherein the determination is made as “navel”.   Detecting the outline of the bean grain based on the density difference of the image data on the surface of the bean grain, determining the presence or absence of flatness based on the ratio of the maximum vertical dimension and the maximum horizontal dimension in the outline, and the surface of the bean grain 3. The bean quality determination method according to claim 1, wherein the determination of the presence / absence of poor endothelial adhesion based on the density difference of the image data is also executed. A bean sorting device that exposes the “navel” for incorporating nutrients for growth on the surface, and determines the quality of the bean grains that are wrapped in the epidermis of a predetermined concentration range. There,
An imager that scans and images the surface of the beans, and performs a plurality of times while relatively moving in a direction orthogonal to the scanning direction;
Analyzing the image data captured by the imaging means, classifying means for classifying the normal epidermal concentration site and the special epidermal concentration site having a predetermined density difference with respect to the normal epidermal concentration site,
A discriminating means for discriminating whether or not the ratio of the distances (intervals) between adjacent parts that are intermittent in the scanning direction among the special display density parts classified by the classifying means belongs to a predetermined range;
By being affirmed by the determination means, the bean grains containing the special epidermis concentration part are specified to have the “navel”, and the negative judgment is made, to the bean grain containing the special epidermis concentration part, A specific means for identifying the presence of a “crack” in which the epidermis has ruptured;
Sorting means for separating defective products of other items including at least beans with “crack” identified by the identifying means and extracted based on the analysis result of the image data imaged by the imaging means;
A bean grain sorting device.

Images