JP4664778B2 - Chestnut quality inspection method - Google Patents

Description

  The present invention relates to a method for inspecting non-destructively and non-contactingly whether or not a fruit from which chestnuts have been removed has an internal defect.

  Chestnut fruit is composed of edible pulp, astringent skin as the inner skin that covers the surface of the pulp, and demon skin as an outer skin that covers the astringent skin from the outside. On the surface of the pulp, there are grooves and irregularities that extend from the head of the outer skin (the pointed point) to the seat (the rough part on the bottom side). The astringent skin bites into the groove, making it a thick stripe.

  As a method for inspecting whether or not the chestnut fruit has an internal defect such as worm-eaten, the fruit is irradiated with X-rays, an X-ray image transmitted through the fruit is imaged, and transmission image information obtained by the imaging The region corresponding to the fruit is identified, and the presence or absence of an internal defect (cavity) in the fruit is determined depending on whether or not there is an isolated region having a density different from the pixel of the region in the region corresponding to the fruit. It is known to judge without destroying (for example, see Patent Document 1).

When X-rays pass through the fruit, the X-ray intensity is attenuated by the interaction with the substance (photoelectric absorption, Compton scattering, etc.). The rate of attenuation is proportional to the distance that X-rays have passed through the fruit. If there is a defect (cavity) inside the fruit, this defect part is more likely to transmit X-rays than the pulp part, so the area corresponding to the defect part in the transmission image information is white or gray on the screen of a monitor or the like. It is displayed in color (with high density pixels). On the other hand, the region corresponding to the fruit in the transmission image information is displayed in black (with low density pixels). The conventional method is to determine the presence or absence of an internal defect of a fruit using such a difference in shade.
JP 2004-181432 A

  However, in the above-described conventional method, when X-rays pass through the stripes of the astringent skin that have digged into the grooves on the pulp surface, the area corresponding to the stripes in the transmitted image information may be displayed white or grayish on the screen of a monitor or the like. is there. For this reason, although there is no defect inside the fruit, there is a possibility that such a streak may be misidentified as an internal defect, and the internal quality of the chestnut fruit cannot be properly inspected.

  Therefore, the present invention has a technical problem to improve the above-described present situation.

  In order to solve this technical problem, the chestnut quality inspection method according to the invention of claim 1 includes an imaging step of irradiating the chestnut fruit with electromagnetic waves and imaging an electromagnetic wave image transmitted through the fruit, and the imaging. A smoothing step of smoothing the transmission image information obtained by a plurality of smoothing filters to identify a fruit region from each of the smoothed image information, and any one of the smoothed fruit regions The difference process of calculating the difference between the smoothed fruit regions other than this and specifying each extracted region extracted by the difference calculation as a primary defect candidate region, and binarizing the transmission image information A binarization step of processing and identifying the fruit region from the binarized image information, and each isolated region of the high density pixel surrounded by the low density pixel in the binarized fruit region is a second defect. As a candidate area, the surface of each secondary defect candidate area If the area of the second defect candidate region corresponding to the first defect candidate region is outside the range of the preset area, the first corresponding to each other. And the second defect candidate area includes a first determination step for determining that the defect area corresponds to an internal defect of the fruit.

  According to a second aspect of the present invention, in the chestnut quality inspection method according to the first aspect, in each of the second defect candidate regions other than the fruit determined to be an internal defect in the first determination step, in the fruit A length measuring step for measuring a vertical length along the vertical axis direction from the head to the seat and a horizontal length along the horizontal axis direction orthogonal to the vertical axis direction, and an aspect ratio of the vertical length to the horizontal length. The aspect ratio calculation step to be obtained and the aspect ratio of the second defect candidate region in the fruit other than the fruit determined to be an internal defect in the first determination step is out of a preset aspect ratio range, or the And a second determination step of determining that the second defect candidate area corresponds to an internal defect of the fruit if the horizontal length of the secondary defect candidate area is greater than a preset horizontal length. something like A.

  According to the chestnut quality inspection method according to the invention of claim 1, after the imaging process → smoothing process → difference process → binarization process → area calculation process, the first determination process obtains the difference process. The second defect candidate corresponding to the first defect candidate area is obtained by grasping the correspondence relationship between the first defect candidate area and the second defect candidate area obtained in the binarization step. If the area of the region is out of the range of the preset set area, the first and second defect candidate regions corresponding to each other are determined to correspond to the internal defect of the fruit. The first and second defect candidate areas having the above-described area, that is, the streak areas appearing in the transmission image information obtained in the imaging process are not mistaken as internal defects. Therefore, it is possible to discriminate between internal defects and astringent lines in the fruit with higher accuracy than in the conventional method, and it is possible to greatly reduce the occurrence of inspection errors (discrimination errors).

  Moreover, according to the chestnut quality inspection method according to the invention of claim 2, for each of the second defect candidate regions, the longitudinal length along the vertical axis direction from the head to the seat in the fruit, and the vertical axis direction After measuring the horizontal length along the direction of the transverse axis perpendicular to each other, and then determining the aspect ratio of the vertical length to the horizontal length, the first in the fruit other than the fruit determined to be an internal defect in the first determination step If the aspect ratio of the secondary defect candidate area is out of a preset set aspect ratio, or if the lateral length of the secondary defect candidate area is larger than the preset set lateral length, the secondary Since it is determined that the defect candidate area corresponds to an internal defect of the fruit, an internal defect that is difficult to find only by the first determination step, that is, an internal defect having the same size (area) as a streak, or near the outer periphery of the fruit Part of Internal defects in, can be found reliably after having distinguished from streaks of astringent skin. Therefore, there is an effect of further improving the accuracy of identifying internal defects and streaks in the fruit.

  DESCRIPTION OF EMBODIMENTS Embodiments embodying the present invention will be described below with reference to the drawings (FIGS. 1 to 25).

  FIG. 1 is a schematic side view showing a chestnut quality inspection apparatus, FIG. 2 is a functional block diagram of a controller as a control means, FIG. 3 is a flowchart showing a procedure for discriminating internal defects of fruits, and FIG. FIGS. 5A to 5C are explanatory diagrams of a screen showing smoothed image information, FIG. 6 is an explanatory diagram of a smoothing process for transparent image information, and FIG. 7A. (B) is an explanatory diagram of a screen showing difference image information, FIG. 8 is an explanatory diagram of a screen showing binarized image information, FIG. 9 is an explanatory diagram of a screen showing labeled image information, and FIG. FIGS. 11A to 11C are explanatory diagrams of screens showing smoothed image information, and FIGS. 12A and 12B are screens showing difference image information. FIG. 13 is an explanatory diagram of a screen showing binarized image information, and FIG. FIG. 15 is an explanatory diagram of a screen showing transparent image information in the third embodiment, and FIGS. 16A to 16C are explanatory diagrams of screens showing smoothed image information. 17A and 17B are explanatory diagrams of a screen showing difference image information, FIG. 18 is an explanatory diagram of a screen showing binarized image information, and FIG. 19 is an explanatory diagram of a screen showing labeled image information. FIG. 20 is an explanatory diagram of a screen showing labeled image information in the fourth embodiment, FIG. 21 is a flowchart showing a determination procedure in the fifth embodiment, and FIG. 22 is an explanatory diagram of a screen showing transparent image information, FIG. 23 is an explanatory diagram of a screen showing binarized image information, FIG. 24 is an explanatory diagram of a screen showing labeled image information, and FIG. 25 is a flowchart showing a procedure of a determination process.

<Outline of quality inspection equipment>
As shown in FIG. 1, the chestnut quality inspection apparatus 1 sandwiches the transport conveyor line 2 that transports the fruits C from which chestnuts are removed in a row at an appropriate pitch interval, and the transport conveyor line 2. A position detection sensor for detecting whether or not a fruit C exists at a place where the electromagnetic wave irradiator 3 and the electromagnetic wave image pickup device 4 arranged so as to face each other at the upper and lower parts and the electromagnetic wave irradiator 3 in the transport conveyor line 2 are present. 5, the blower 6 disposed on the downstream side of the electromagnetic wave irradiator 3 and the electromagnetic wave image pickup device 4 in the transfer conveyor line 2, and the drive of the transfer conveyor line 2, the electromagnetic wave irradiation device 3, the electromagnetic wave image pickup device 4, and A controller 7 is provided as control means for executing various image processing based on control data from the position detection sensor 5.

  In this example, the X-ray irradiator 13 that irradiates the fruit C on the transport conveyor line 2 positioned below the X-ray that is one of the electromagnetic waves is employed as the electromagnetic wave irradiator 3. As the electromagnetic wave image pickup device 4, an X-ray image pickup device 14 that picks up an X-ray image transmitted through the fruit C and outputs transmission image information obtained by the image pickup to the controller 7 is employed. The position detection sensor 5 is a proximity sensor such as light (infrared ray), ultrasonic wave, or capacitive type.

  In the quality inspection apparatus 1 of the embodiment, when the position detection sensor 5 senses that the fruit C has arrived below the X-ray irradiator 13 by driving the transport conveyor line 2, the X-ray irradiator 13 The X-ray image which permeate | transmitted this fruit C by irradiating X-ray | X_line toward the downward fruit C is imaged with the X-ray imaging device 14, The transmission image information (refer FIG. 4) obtained by this imaging is X Transmission from the line imager 14 to the controller 7 is performed. And the said quality inspection apparatus 1 is comprised so that the fruit C which the controller 7 judged to have an internal defect as a result of the image process mentioned later may be removed from the conveyance conveyor line 2 with the selection wind from the air blower 6. .

  The controller 7 includes a central processing unit 8 (CPU) that executes various operations for image processing and the like, a read only memory 9 (ROM) that stores a control program, and an arbitrarily read / writeable memory 10 that stores various data. (RAM), a timer for timing, an input / output interface, and the like (see FIG. 2).

  The input / output interface of the controller 7 includes the X-ray irradiator 13, the X-ray imager 14, the position detection sensor 5 and the blower 6, a conveyor drive circuit 11 for driving the conveyor belt 2, and a quality inspection device. 1 is connected to an input device 12 having a main switch, a keyboard and the like for turning on and off the entire power supply, and a display device 15 such as a liquid crystal monitor.

<Description of internal defect determination procedure>
Next, an example of a procedure (image processing) for determining an internal defect of the fruit C using the above-described quality inspection apparatus 1 will be described.

  As shown in the flowchart of FIG. 3A, the procedure for determining the internal defect of the fruit C in the first embodiment is as follows: imaging step (step S1) → smoothing step (step S2) → difference step (step S3) → second. It is executed in the order of the valuation process (step S4) → the area calculation process (step S5) → the first determination process (step S6).

<First Embodiment>
First, a first embodiment for inspecting a fruit C having two internal defects near the center will be described with reference to FIGS.

  [Imaging Process] In the imaging process of step S1, when the fruit C arrives below the X-ray irradiator 13 by driving the conveyor belt 2, the position detection sensor 5 senses this fact, and based on the detected data. The X-ray irradiator 13 irradiates the fruit C with X-rays. Then, the X-ray imager 14 located below the fruit C captures an X-ray image transmitted through the fruit C. Transmission image information obtained by this imaging is transmitted from the X-ray imaging device 14 to the controller 7 and stored in the RAM 10 in the controller 7.

  By the way, each fruit C on the transport conveyor line 2 lies down in a state where its body part is in contact with the upper surface of the transport conveyor line 2. And the direction of the head (part with a sharp tip) in each fruit C is scattered in all directions. Therefore, on the screen of the display 15, the direction from the head to the seat in the fruit C is defined as the Y-axis (vertical axis) direction, and the direction orthogonal to the Y-axis (vertical axis) direction is defined as the X coordinate. The image information regarding the fruit C is corrected so that is positioned on the upper side of the screen.

  FIG. 4 is an explanatory diagram of a screen showing the transmission image information of the fruit C on the display 15. A hatched portion in the figure indicates a region 17 (hereinafter referred to as a fruit region) corresponding to the fruit C. In the fruit region 17, a vertically long portion substantially along the Y-axis (vertical axis) direction surrounded by a dotted line indicates a region 18 (hereinafter, referred to as a streak region) corresponding to the astringent skin streak in the fruit C. ing. Other white portions indicate a region 19 (hereinafter referred to as a defective region) corresponding to an actual internal defect.

  For convenience of explanation, in the drawing, reference numerals a, b, and c are attached to the stripe region 18 in order from the left side of the drawing. Similarly, the defect area 19 is provided with symbols a and b in order from the left side of the drawing.

  The transmitted image information is represented by shading on the screen of the display 15. In this case, the fruit region 17 filled with pulp is less likely to transmit X-rays compared to the streak region 18 and the defective region 19, so that the fruit region 17 is displayed darkly on the screen of the display 15. The streak area 18 and the defect area 19 are displayed in white or grayish at a higher density than the fruit area 17.

[Smoothing process]
In the smoothing process of step S2, a moving average which is an example of a smoothing process using a plurality of (three in the embodiment) smoothing filters M is applied to the transmission image information stored in the RAM 10 in the controller 7. By performing the processing, a plurality (three in the embodiment) of smoothed image information is created, and a smoothed fruit region 21 (hereinafter referred to as a smoothed fruit region) is specified from each smoothed image information ( (Refer FIG. 5 (a)-(c)).

  Here, the moving average process refers to, for example, a process of replacing the density of each pixel constituting the transmission image information with the average density of the smoothing filter M of N × N pixels centering on an arbitrary pixel PI. (See FIG. 6, FIG. 6 shows a 3 × 3 pixel smoothing filter M). The smoothing filter M of the first embodiment is a simple average filter that replaces the density of each pixel with the average density of the smoothing filter M of N × N pixels centered on an arbitrary pixel PI. In addition, a median filter, a maximum / minimum value filter, an opening / closing filter, or the like may be employed.

  FIGS. 5A to 5C are explanatory diagrams of screens showing smoothed image information. The three smoothing filters M used in the first embodiment have different values of the order N. As the value of the order N is larger, the filter table becomes larger and the smoothness of the smoothed image information becomes higher (noise becomes smaller), but the entire smoothed image information tends to become soft (become blurred). is there.

  The smoothing filter M having a large value of the order N is effective for smoothing and extracting a large size (area) defect region 19 (including the streak region 18) from the transmission image information. 19 is difficult to extract (easily lost from the smoothed image information). On the contrary, the smoothing filter M having a small value of the order N is effective for smoothing and extracting the defect area 19 having a small size from the transmission image information, but is difficult to extract the defect area 19 having a large size. There is.

  The order N of the smoothing filter M in each smoothed image information is N = N1, N2, N3 (N1 to N3 are all integers) in alphabetical order attached to the figure numbers. These orders have a relationship of N1> N2> N3. Accordingly, although not shown in detail, the smoothed image information shown in the screens of FIGS. 5A to 5C has a noise that decreases in the order of (c) → (b) → (a), and there is a change in density. It is smooth.

  Further, as is clear from the relationship between the orders N of the smoothing filter M, the screen of FIG. 5C has a smoothed defect area 23a and a smoothed streak area 22a of a large size in the smoothed fruit area 21c. Is displayed. On the screen of FIG. 5B, a smoothed defect area 23b and smoothed streak areas 22b and 22c having a small size are displayed in the smoothed fruit area 21b. The screen of FIG. 5A displays a smoothed fruit region 21a that has been smoothed to a high density pixel and has lost both the smoothed defect region 23 and the smoothed streak region 22 because the order N = N1 is too large. Has been. The screen of FIG. 5A is the same as the state in which the smoothed image information of the normal fruit C without a cavity is displayed.

[Difference process]
In the difference step of step S3, the difference between any one smoothed fruit region 21 (smoothed fruit region 21a in FIG. 5A in the first embodiment) and the other smoothed fruit regions 21 is calculated. By doing this, the number of pieces of difference image information that is one less than the number of pieces of smoothed image information (three in the first embodiment) is created, and the extracted area (the extracted region extracted by the difference calculation) among these pieces of difference image information ( 7 (a) and 7 (b) are identified as primary defect candidate regions 24 that may correspond to actual internal defects.

  In the first embodiment, by subtracting the smoothed fruit region 21b shown in FIG. 5 (b) from the smoothed fruit region 21a shown in FIG. 5 (a), the difference image information shown in the screen of FIG. 7 (a) is obtained. can get. In the difference image information corresponding to FIG. 7A, only the region having a large difference value (density difference), that is, the region corresponding to the smoothing defect region 23b and the smoothing streak regions 22b and 22c remains extracted. . These extraction areas are specified as primary defect candidate areas 24c, 24d, and 24e.

  Moreover, the difference image information shown in the screen of FIG. 7B is obtained by subtracting the smoothed fruit region 21c shown in FIG. 5C from the smoothed fruit region 21a shown in FIG. From the difference image information corresponding to FIG. 7B, areas corresponding to the smoothed defect area 23a and the smoothed streak area 22a are extracted and specified as the primary defect candidate areas 24a and 24b.

[Binarization process]
In the binarization step of step S4, the binarization process shown in the screen of FIG. 8 is performed by performing binarization processing on the transmission image information shown in the screen of FIG. 4 using a predetermined threshold set in advance. Image information is created, and from this binarized image information, a binarized fruit region 25 (hereinafter referred to as binarized fruit region) is specified and surrounded by low density pixels in the binarized fruit region 25. Further, each isolated region of the high-density pixel (see the white area in FIG. 8) is specified as a secondary defect candidate region 26 that may correspond to an actual internal defect. Here, the threshold value used for the binarization process is a small value such that the streak areas 18a to 18c and the like in the transmission image information are not lost after the binarization process using the P tile method, the difference histogram method, or the like. To do.

  In FIG. 8, the black painted portion (low density pixel portion) in the figure indicates the binarized fruit region 25. A white isolated area surrounded by a black portion in the binarized fruit area 25 is a secondary defect candidate area 26. Note that the secondary defect candidate region 26 shown in FIG. 8 is appended with symbols a to e in order from the left side of the drawing.

  After the binarization process, low density pixels (black areas) constituting the binarized fruit area 25 and high density pixels (white areas) of the secondary defect candidate area 26 surrounded by the low density pixels. ) And a labeling process to which predetermined symbols are attached, so that the respective regions 25 and 26 are discriminated on the binarized image information (see FIG. 9). In the first embodiment, the label “0” is assigned to the low density pixels constituting the binarized fruit area 25, and the high density pixels in the secondary defect candidate area 26 surrounded by the low density pixels are illustrated. The labels “1”, “2”, “3”, “4”, and “5” are attached in order from the left side of FIG.

[Area calculation process]
In the area calculation step in step S5, the area A of each secondary defect candidate region 26 labeled in the labeling process described above is calculated. The area A of each secondary defect candidate region 26 is obtained by adding the number of pixels occupied by the same label. For convenience of explanation, the area A of the secondary defect candidate region 26 shown in FIG. 9 is appended with symbols α1, α2, α3, α4, and α5 in order from the left side of the drawing.

[First judgment step]
In the first determination step of step S6, first, the binarized image information (see FIG. 9) subjected to the labeling process is compared with each difference image information (see FIGS. 7A and 7B). For each secondary defect candidate area 26 (see FIG. 9), it is determined whether or not there is a corresponding primary defect candidate area 24. As is apparent from FIGS. 7A, 7B, and 9, the secondary defect candidate regions 26 of the first embodiment have primary defect candidate regions 24 corresponding to the five locations, respectively. . That is, the secondary defect candidate region 26a in FIG. 9 corresponds to the primary defect candidate region 24a in FIG. 7B, and the secondary defect candidate region 26b in FIG. 9 is the primary defect candidate region 26b in FIG. 7B. 9 corresponds to the defect candidate region 24b, the second defect candidate region 26c in FIG. 9 corresponds to the first defect candidate region 24c in FIG. 7A, and the second defect candidate region 26d in FIG. Corresponding to the primary defect candidate region 24d of a), the secondary defect candidate region 26e of FIG. 9 corresponds to the primary defect candidate region 24e of FIG. 7A.

  Next, it is determined whether or not the area A of the secondary defect candidate region 26 having the corresponding primary defect candidate region 24 is outside the preset set areas Smin to Smax. The range of the set areas Smin to Smax assumes the area of streaks of fruit skin C in the fruit C, and the portions corresponding to the first and second defect candidate areas 24 and 26 corresponding to each other in the fruit C are streaks of the astringent skin. It is set as a reference for determining whether or not there is.

  In the first embodiment, the areas Aα2, Aα4, and Aα5 of the three vertically long secondary defect candidate regions 26b, 26d, and 26e are the set minimum area Smin or more and the set maximum area Smax or less (Smin ≦ Aα2, Aα4, respectively). Since Aα5 ≦ Smax), it is determined that the portions corresponding to the secondary defect candidate areas 26b, 26d, and 26e (primary defect candidate areas 24b, 24d, and 24e) in the fruit C are streaks and are not internal defects. Is done.

  On the other hand, since the area Aα1 of the secondary defect candidate region 26a is larger than the set maximum area Smax (Smax <Aα1), the portion corresponding to the secondary defect candidate region 26a (first defect candidate region 24a) in the fruit C is Judged as an internal defect. Further, since the area Aα3 of the secondary defect candidate region 26c is smaller than the set minimum area Smin (Smin> Aα3), the portion corresponding to the secondary defect candidate region 26c (first defect candidate region 24b) in the fruit C is Judged as an internal defect. Then return. Even if it is determined that the portions corresponding to the first and second defect candidate regions 24 and 26 corresponding to each other in the fruit C are all streaks and are not internal defects, the process returns as it is.

  As a result of the image processing by the controller 7, when it is determined that the fruit C below the X-ray irradiator 13 has an internal defect, an appropriate time is required after the position detection sensor 5 senses the arrival of the fruit C. After the elapse of (the time required to convey the fruit C forward of the blower 6), the blower 6 is driven. And the fruit C with the internal defect which arrived ahead of the air blower 6 is removed from the conveyance conveyor line 2 by the selection wind from this air blower 6 (refer FIG. 1). When it is determined that there is no internal defect in the fruit C below the X-ray irradiator 13, the fruit C is directly placed on the transport conveyor line 2 and sent to the next process such as bagging.

<Operational effects of the first embodiment>
According to such a method, if the area A of the secondary defect candidate region 26 corresponding to the primary defect candidate region 24 is outside the preset set areas Smin to Smax, the secondary in the fruit C Since it is determined that the portion corresponding to the defect candidate region 26 (defect candidate region 24) is an internal defect, the second defect candidate region 26b having areas Aα2, Aα4, Aα5 within the range of the set areas Smin to Smax, 26d, 26e (first defect candidate areas 24b, 24d, 24e), that is, the streak areas 18a to 18c appearing in the transmission image information (see FIG. 4) are not mistaken as internal defects. Therefore, it is possible to discriminate internal defects and astringent skin stripes in the fruit C with higher accuracy than in the conventional method, and the occurrence of inspection errors (discrimination errors) can be greatly reduced.

<Second Embodiment>
Next, with reference mainly to FIGS. 10 to 14, a second embodiment for inspecting a fruit C having an internal defect of the same size (area) as a streak in the vicinity of the center will be described. The second embodiment shows an example of finding an internal defect having a size (area) comparable to a streak among internal defects that cannot be found only by the first determination step.

[Imaging process-Area calculation process]
As shown in FIG. 3B, in the second embodiment, the procedure up to the first determination step in step S6 ′ is basically the same as that in the first embodiment. After the first determination step, whether to return as it is depending on the correspondence relationship between the first defect candidate region 34 (extraction region) and the second defect candidate region 36 (isolated region), which will be described later, and the presence or absence of internal defects. The length measurement process (step S7 ′) → the aspect ratio calculation process (step S8 ′) → the second determination process (step S9 ′).

  FIG. 10 is an explanatory diagram of a screen showing the transmitted image information of the fruit C obtained by imaging. The hatched portion in the figure indicates the fruit region 27. Of the white portions surrounded by dotted lines in the fruit region 27, those vertically long along the Y-axis (vertical axis) direction are streak regions 28a to 28c, and the other white portions are defect regions 29a.

  FIGS. 11A to 11C are explanatory diagrams of screens showing smoothed image information. In the screen of FIG. 11 (c), a large-sized smoothing defect region 33a and a smoothing stripe region 32a are displayed in the smoothed fruit region 31c, and the smoothed fruit region is displayed on the screen of FIG. 11 (b). The smoothed streak areas 32b and 32c having a small size are displayed in 31b. On the screen of FIG. 11A, a smoothed fruit region 31a that has been smoothed to a high density pixel and has lost both the smoothed defect region 33 and the smoothed streak region 32 is displayed.

  FIGS. 12A and 12B are explanatory diagrams of screens showing difference image information. In the difference image information shown on the screen of FIG. 12A, only the regions corresponding to the smoothed streak regions 32b and 32c are extracted and remain. These extraction areas are specified as primary defect candidate areas 34c and 34d. In the difference image information shown in the screen of FIG. 12B, only the areas corresponding to the smoothed defect area 33a and the smoothed streak area 32a are extracted and remain. These extraction areas are also specified as the primary defect candidate areas 34a and 34b.

  FIG. 13 is an explanatory diagram of a screen showing binarized image information. In FIG. 13, black portions (low density pixel portions) in the figure indicate the binarized fruit region 35. White isolated regions surrounded by black portions in the binarized fruit region 35 indicate secondary defect candidate regions 36a to 36d. In addition, the areas 36b to 36d corresponding to the stripe regions 28a to 28c among the secondary defect candidate regions 36 shown in the screen of FIG. 14 are displayed as Aβ2, Aβ3, and Aβ4. The area of the object 36a corresponding to the defect region 29a is indicated as Aβ1.

[First judgment step]
In the first determination step of step S6 ′, the binarized image information (see FIG. 13) subjected to the labeling process is compared with each difference image information (see FIGS. 12 (a) and 12 (b)). It is determined for each secondary defect candidate area 36 whether or not there is a corresponding primary defect candidate area 34.

  Since all the four secondary defect candidate regions 36 of the second embodiment have corresponding primary defect candidate regions 34, the second defect candidate regions 34 having the corresponding primary defect candidate regions 34 are then included. It is determined whether or not the area A of the next defect candidate region 36 is outside a preset set area Smin to Smax.

  In the second embodiment, since the areas Aβ1 to Aβ4 of the secondary defect candidate regions 36a to 36d are all equal to or larger than the set minimum area Smin and equal to or smaller than the set maximum area Smax (Smin ≦ Aβ1 to Aβ4 ≦ Smax), The portions corresponding to the primary and secondary defect candidate regions 34 and 36 corresponding to each other are determined to be astringent skin streaks and not internal defects, and the process proceeds to the next length measurement step (step S7 ').

[Length measurement process]
In the length measuring step in step S7 ′, the vertical length L in the Y-axis (vertical axis) direction and the horizontal length H in the X-axis (horizontal axis) direction perpendicular to the Y-axis direction are obtained for the secondary defect candidate region 36. And measure. As shown in FIG. 14, the vertical length L and the horizontal length H of each secondary defect candidate region 36 are displayed with the symbols βa, βb, βc, βd sequentially from the left side of the drawing.

[Aspect ratio calculation process]
Next, in the aspect ratio calculation step of step S8 ′, the aspect ratio L / H of the vertical length L to the horizontal length H in each secondary defect candidate region 36 is obtained, and the process proceeds to the next second determination step (step S9). .

[Second determination step]
In the second determination step of step S9 ′, it is determined whether or not the aspect ratio L / H of each secondary defect candidate region 36 is outside the preset set aspect ratios Fmin to Fmax. The range of the set aspect ratio Fmin to Fmax assumes the aspect ratio of astringent skin streaks in the fruit C.

  In the second embodiment, the aspect ratios Lβb / Hβb, Lβc / Hβc, and Lβd / Hβd of the three vertically long secondary defect candidate regions 36b to 36d are not less than the set minimum aspect ratio Fmin and not more than the set maximum aspect ratio Fmax ( Fmin ≦ Lβb / Hβb, Lβc / Hβc, Lβd / Hβd ≦ Fmax), so that it is excluded from the determination of whether or not there is an internal defect.

  Since the aspect ratio Lβa / Hβa of the remaining secondary defect candidate area 36a is smaller than the set minimum aspect ratio Fmin (Fmin> Lβa / Hβa), the primary and secondary defect candidate areas 34a, A portion corresponding to 36a is determined to be an internal defect. Then return. The fruit C determined to have an internal defect is removed from the conveying conveyor line 2 by the sorting air from the blower 6 as in the case of the first embodiment.

  If it is determined that all the portions corresponding to the primary and secondary defect candidate areas 34a and 36a corresponding to each other in the fruit C are streaks of astringent skin and are not internal defects, lateral length determination described later is performed. (Details will be described in the fourth embodiment).

<Third Embodiment>
Next, a third embodiment for inspecting a fruit C having two internal defects at a portion near the outer periphery will be described with reference to FIGS. 3rd Embodiment has shown the example in the case of finding the internal defect in the site | part near the outer periphery in the fruit C among the internal defects which cannot be found only by a 1st judgment process.

[Imaging process-Area calculation process]
Also in 3rd Embodiment, the discrimination | determination procedure of the internal defect in the fruit C is performed similarly to 2nd Embodiment along the flowchart shown in FIG.3 (b).

  FIG. 15 is an explanatory diagram of a screen showing transmission image information of the fruit C obtained by imaging. The hatched portion in the figure indicates the fruit region 37. Of the white portions surrounded by dotted lines in the fruit region 37, those vertically long along the Y-axis (vertical axis) direction are streak regions 38a to 38c, and the other white portions are defect regions 39a and 39b. .

  FIGS. 16A to 16C are explanatory diagrams of screens showing smoothed image information. In the screen of FIG. 16 (c), a large size smoothed stripe region 32a is displayed in the smoothed fruit region 41c, and in the screen of FIG. 16 (b), a small size is displayed in the smoothed fruit region 41b. Although the smoothing streak areas 42b and 42c are displayed, two smoothing defect areas do not appear. This is because the defect areas 39a and 39b are located near the outer periphery of the fruit C, and the difference in density between the defect areas 39a and 39b and the fruit area 37 in the transmission image information (see FIG. 15) is originally small. This is because it is smoothed by the smoothing process of ′.

  For this reason, in the difference process of step S3 ′, the difference between the smoothed fruit region 41a of FIG. 16A and the smoothed fruit regions 41b and 41c of FIG. 16B and FIG. In the difference image information (see FIGS. 17A and 17B), only the areas corresponding to the smoothed streak areas 42a or 42b and 42c having a large difference value (density difference) are extracted and remain. These extraction areas are specified as the primary defect candidate areas 44b or 44d, 44e.

  FIG. 18 is an explanatory diagram of a screen showing binarized image information. In FIG. 18, black portions (low-density pixel portions) in the figure indicate the binarized fruit region 45. White isolated regions surrounded by black portions in the binarized fruit region 45 indicate second defect candidate regions 46a to 46e. In addition, the areas 46b, 46d, and 46e corresponding to the stripe regions 38a to 38c among the secondary defect candidate regions 46 shown in the screen of FIG. 19 are displayed as Aγ2, Aγ4, and Aγ5. Areas 46a and 46c corresponding to the defect regions 39a and 39c are indicated as Aγ1 and Aγ3.

[First judgment step]
In the first determination step of step S6 ′, the binarized image information (see FIG. 18) that has been subjected to the labeling process is compared with each difference image information (see FIGS. 17A and 17B). It is determined for each secondary defect candidate area 46 whether or not there is a corresponding primary defect candidate area 44.

  In the third embodiment, out of the five secondary defect candidate regions 46, there are corresponding primary defect candidate regions for the two things 46a and 46c located near the outer periphery of the binarized fruit region 45. Although not, there are primary defect candidate regions 44b, 44d, and 44e corresponding to the three vertically long secondary defect candidate regions 46b, 46d, and 46e.

  Therefore, it is determined whether or not the area A of the secondary defect candidate region 46 having the corresponding primary defect candidate region 44 is outside the preset set areas Smin to Smax.

  In the third embodiment, the areas Aγ2, Aγ4, and Aγ5 of the three vertically long secondary defect candidate regions 46b, 46d, and 46e are the set minimum area Smin and the set maximum area Smax (Smin ≦ Aγ2, Aγ4, respectively). Since Aγ5 ≦ Smax), the portions corresponding to the second defect candidate regions 46b, 46d, and 46e (first defect candidate regions 44b, 44d, and 44e) in the fruit C are determined to be astringent skin streaks and not internal defects. Then, the process proceeds to the next length measurement step (step S7 ′).

[Length measurement process]
In the length measuring step in step S7 ′, the vertical length L in the Y-axis (vertical axis) direction and the horizontal length H in the X-axis (horizontal axis) direction orthogonal to the Y-axis direction are obtained for the secondary defect candidate region 46. And measure. As shown in FIG. 19, the vertical length L and the horizontal length H of each secondary defect candidate region 46 are displayed with reference numerals γa, γb, γc, and γd in order from the left side of the drawing.

[Aspect ratio calculation process]
Next, in the aspect ratio calculation step of step S8 ′, the aspect ratio L / H of the vertical length L to the horizontal length H in each secondary defect candidate region 46 is obtained, and the process proceeds to the next second determination step (step S9 ′). To do.

[Second determination step]
In the second determination step of step S9 ′, it is determined whether or not the aspect ratio L / H of each secondary defect candidate area 46 is outside the preset set aspect ratios Fmin to Fmax.

  In the second embodiment, the aspect ratios Lγb / Hγb, Lγc / Hγc, and Lγd / Hγd of the three vertically long secondary defect candidate regions 46b to 46d are not less than the set minimum aspect ratio Fmin and not more than the set maximum aspect ratio Fmax ( Fmin ≦ Lγb / Hγb, Lγc / Hγc, Lγd / Hγd ≦ Fmax), so that it is excluded from the determination of whether or not there is an internal defect.

  In addition, since the aspect ratio Lγa / Hγa of the secondary defect candidate region 46a is also not less than the set minimum aspect ratio Fmin and not more than the set maximum aspect ratio Fmax (Fmin ≦ Lγa / Hγa ≦ Fmax), it is an object to determine whether it is an internal defect or not. It is assumed to be outside.

  Since the aspect ratio Lγc / Hγc of the remaining secondary defect candidate area 46c is smaller than the set minimum aspect ratio Fmin (Fmin> Lγc / Hγc), the portion corresponding to the secondary defect candidate area 46c in the fruit C is an internal defect. It is judged that there is. Then return. The fruit C determined to have an internal defect is removed from the conveying conveyor line 2 by the sorting air from the blower 6 as in the case of the first embodiment.

  If it is determined that both of the portions corresponding to the two secondary defect candidate regions 46a and 46c in the fruit C are streaks of streaks and are not internal defects, lateral length determination described later is performed (details are given below) This will be described in the fourth embodiment).

<Fourth embodiment>
FIG. 20 is a fourth embodiment in the case where there is no primary defect candidate area corresponding to the secondary defect candidate area. The fourth embodiment shows an example in which there are three streak regions inside the fruit C and there is one internal defect in a portion near the outer periphery.

[Aspect Ratio Discrimination Processing from Imaging Step to Second Determination Step]
In the fourth embodiment, in the second determination step of step S9 ′, the aspect ratio L / H of the secondary defect candidate areas 56 in all four places is not less than the set minimum aspect ratio Fmin and the set maximum aspect ratio. Since the ratio is equal to or less than Fmax (Fmin ≦ L / H ≦ Fmax), the process is executed in the same manner as in the second and third embodiments up to the point where it is excluded from the determination of whether or not there is an internal defect (see FIG. 3B). ).

  FIG. 20 is an explanatory diagram of a screen showing image information after the labeling process. The area labeled “0” in the screen of FIG. 20 indicates the low density pixel (black pixel) portion that constitutes the binarized fruit area 55. The region labeled “1” is the second defect candidate region 56a corresponding to the defect region of the fruit C. The regions labeled “2”, “3”, and “4” are secondary defect candidate regions 56b, 56c, and 56d corresponding to the streak region of the fruit C. Note that, for convenience of explanation, the area A of the secondary defect candidate region 56 shown in FIG.

[Horizontal length discrimination processing]
After the aspect ratio determination process in the second determination step of step S9 ′, it is determined whether or not the horizontal length H of each secondary defect candidate region 56 is larger than a preset horizontal length HO. The set horizontal length HO assumes the maximum value of the horizontal length of the stripes in the fruit C.

  In the fourth embodiment, since the horizontal lengths Hδb to Hδd of the three vertically long secondary defect candidate regions 56b to 56d are equal to or less than the set horizontal length HO (HO ≧ Hδb to Hδd), it is not subject to determination as to whether or not they are internal defects. It is said. Since the horizontal length Hδa of the remaining secondary defect candidate region 56a is larger than the set horizontal length HO (HO <Hδa), it is determined that the portion corresponding to the secondary defect candidate region 56a in the fruit C is an internal defect, Return. The fruit C determined to have an internal defect is removed from the transport conveyor line 2 by the sorting air from the blower 6 as in the first and second embodiments.

  Even if it is determined that all the portions corresponding to the secondary defect candidate area 56 in the fruit C are not internal defects, the process directly returns. And the said fruit C will be mounted on the conveyance conveyor line 2 as it is, and will be sent to subsequent processes, such as bagging.

<Operational effects of the second to fourth embodiments>
As described above, according to the methods shown in the second to fourth embodiments, the aspect ratio L / H is outside the range of the preset aspect ratios Fmin to Fmax, or the lateral length H is preset. If it is larger than the set lateral length HO, it is determined that the portion corresponding to the second defect candidate areas 36, 46, 56 in the fruit C is an internal defect. Therefore, it is found only by the first determination process in step S6 ′. A hard defect area, that is, an internal defect having the same size (area) as a streak or an internal defect near the outer periphery of the fruit C can be surely found after distinguishing from a streak skin streak. Therefore, the accuracy of identifying internal defects in fruit C and streaks of astringency is further improved.

<Fifth Embodiment>
21 to 25 show a fifth embodiment in the case of adopting a determination procedure different from that of the first to fourth embodiments described above. As shown in the flowchart of FIG. 21, the procedure for determining the internal defect of the fruit C in the fifth embodiment is an imaging step (step T1) → binarization step (step T2) → area calculation step (step T3) → length. The measurement process (step T4) → the aspect ratio calculation process (step T5) → the determination process (step T6) is executed. Generally speaking, the difference from the first to fourth embodiments is that there are no two steps, a smoothing step and a difference step.

[Imaging process-Aspect ratio calculation process]
In the fifth embodiment, the imaging process at step T1, the binarization process at step T2, the area calculation process at step T3, the length measurement process at step T4, and the aspect ratio calculation process at step T5 are all first. ~ Performed in the same manner as the corresponding steps of the fourth embodiment.

  FIG. 22 is an explanatory diagram of a screen showing transmission image information of the fruit C obtained by imaging. The shaded area in the figure is the fruit region 57. Of the white portions surrounded by dotted lines in the fruit region 57, those vertically long along the Y-axis (vertical axis) direction are streak regions 58a to 58c, and the other white portions are defect regions 59a and 59b. .

  FIG. 23 is an explanatory diagram of a screen showing binarized image information. In FIG. 23, the blackened portion (low density pixel portion) in the figure shows the binarized fruit region 65. White portions surrounded by black portions in the binarized fruit region 65 indicate defect candidate regions 66a to 66e which are isolated regions. In addition, areas 66a, 66b, and 66d corresponding to the stripe regions 58a to 58c among the defect candidate regions 66 shown in the screen of FIG. 24 are displayed as A1 ', A2', and A4 '. The areas of the portions 66c and 66e corresponding to the defect region 59a are indicated as A3 'and A5'.

[Judgment process]
In the determination step of Step T6 in the fifth embodiment, for example, it is executed according to the flowchart shown in FIG.

  First, it is determined whether or not the area A ′ of the defect candidate region 66 shown in FIG. 22 is outside the preset set areas Smin to Smax (step T61). Also in this case, the range of the set areas Smin to Smax assumes the area of streaks on the fruit C.

  In the fifth embodiment, the areas A1 ′, A2 ′, A4 ′ of the three vertically long defect candidate regions 66a, 66b, 66d are all not less than the set minimum area Smin and not more than the set maximum area Smax (Smin ≦ A1 ′, A2). ′, A4 ′ ≦ Smax), the portions corresponding to the defect candidate regions 66a, 66b, 66d in the fruit C are determined not to be internal defects at this stage.

  On the other hand, since the area A3 ′ of the defect candidate area 66c is smaller than the set minimum area Smin (Smin> A3 ′), it is determined that the portion corresponding to the defect candidate area 66c in the fruit C is an internal defect. In addition, since the area A5 ′ of the other defect candidate region 66e is larger than the set maximum area Smax (Smax <A5 ′), it is determined that the portion corresponding to the defect candidate region 66e in the fruit C is also an internal defect.

  Therefore, since it is determined that at least one defect candidate area 66 is an internal defect (T61: YES), it is determined in step T65 that the fruit C to be determined has an internal defect, and then the process returns.

  When the area A ′ of all defect candidate regions 66 is not less than the set minimum area Smin and not more than the set maximum area Smax (Smin ≦ A ′ ≦ Smax) (T61: NO), then the aspect ratio of each defect candidate region 66 is It is determined whether or not l / h is out of a preset set aspect ratio Fmin to Fmax (step T62). The range of the set aspect ratio Fmin to Fmax assumes the aspect ratio of astringent skin streaks in the fruit C.

  When the aspect ratio 1 / h of the defect candidate area 66 is outside the range of the set aspect ratio Fmin to Fmax (T62: YES), it is determined that the portion corresponding to the defect candidate area 66 in the fruit C is an internal defect, In step T65, the fruit C to be determined is recognized as having an internal defect, and then the process returns.

  When the aspect ratio 1 / h of all defect candidate areas 66 is not less than the set minimum aspect ratio Fmin and not more than the set maximum aspect ratio Fmax (Fmin ≦ l / h ≦ Fmax) (T62: NO), then the defect candidate area It is determined whether or not the horizontal length h of 46 is larger than a preset horizontal length HO (step T63). The set horizontal length HO assumes the maximum value of the horizontal length of the stripes in the fruit C.

  When the lateral length h of the defect candidate area 66 is greater than the set lateral length HO (T63: YES), the portion corresponding to the defect candidate area 66 in the fruit C is determined to be an internal defect, and the fruit C to be determined is determined in step T65. Is recognized as having an internal defect, and then returns.

  When the horizontal length h of all the defect candidate areas 66 is equal to or less than the set horizontal length HO (HO ≧ h) (T63: NO), the portions corresponding to the defect candidate areas 66 in the fruit C are all astringent skin streaks. In step T64, it is determined that there is no internal defect, and then the process returns.

  The fruit C determined to have an internal defect is removed from the conveying conveyor line 2 by the sorting air from the blower 6 as in the first to fourth embodiments. The fruit C determined to have no internal defect is placed on the transport conveyor line 2 as it is and sent to the next process such as bagging.

<Operational Effects of Fifth Embodiment>
As described above, according to the method shown in the fifth embodiment, the area A ′ of the defect candidate region 66 is outside the preset set areas Smin to Smax, or the aspect ratio l / h is preset. If the set aspect ratio is outside the range of Fmin to Fmax, or if the horizontal length h is larger than the preset horizontal length HO, it is determined that the portion corresponding to the defect candidate region 66 in the fruit C is an internal defect. Therefore, in this case as well, in the same manner as in the first to fourth embodiments, the inside of the fruit C is determined using three judgment criteria: the area A ′ of the defect candidate region 66, the aspect ratio 1 / h, and the horizontal length h. It is possible to discriminate between defects and streaks of skin with extremely high accuracy compared to conventional methods.

  Therefore, also in the case of 5th Embodiment, the internal defect of the fruit C can be reliably found after distinguishing from the streaks of the astringent skin, and the occurrence of inspection errors (discrimination errors) can be greatly reduced.

  Also, in the fifth embodiment, unlike the first to fourth embodiments, two steps of the smoothing step and the difference step are not required, so the total number of steps can be reduced and the efficiency of the quality inspection work is improved. Can also contribute.

<Others>
The present invention is not limited to the above-described embodiment, and can be embodied in various forms. For example, in the fifth embodiment, the area calculation process in step T3 is eliminated, and the presence or absence of internal defects in the fruit C is determined based on two judgment criteria of the aspect ratio l / h and the horizontal length h of the defect candidate region 66. You may make it discriminate | determine after distinguishing from a stripe. Also in this case, the streak area appearing in the transmission image information obtained in the imaging process in step T1 is not mistaken for an internal defect, and the internal defect and the astringent skin streak in the fruit C can be distinguished with high accuracy.

  However, in the fifth embodiment, the accuracy of discriminating internal defects and astrings of astringent skin in the fruit C is determined based on three determination criteria: the area A ′ of the defect candidate region 66, the aspect ratio 1 / h, and the horizontal length h. Needless to say, it is more accurate.

  An electromagnetic wave is a physical electromagnetic wave that includes all those capable of transmitting through an object. In addition to so-called radiation, X-rays, etc., visible light, invisible light, infrared light, far infrared light, ultraviolet light, etc. Is included.

  You may add the process of calculating the gravity center position in the fruit area | regions 25, 35, 45, 55, 65 after a labeling process and before a 1st judgment process. In this case, instead of using the detection result of the position detection sensor 5 as a reference, the blowing timing of the blower 6 can be determined based on the barycentric positions in the regions 25, 35, 45, 55, and 65 of the fruit C. The barycentric positions in the fruit regions 25, 35, 45, 55 and 65 are obtained by adding the pixels occupied by the same label in the horizontal axis (X axis) direction and dividing the added value by the number of pixel columns in the X axis direction. Calculate the center in the X-axis direction, add the pixels occupied by the same label in the vertical axis (Y-axis) direction, and divide the added value by the number of pixel columns in the Y-axis direction. It is obtained by calculating.

  In addition, the configuration of each unit is not limited to the illustrated embodiment, and various modifications can be made without departing from the spirit of the present invention.

It is a schematic side view which shows the quality inspection apparatus of a chestnut. It is a functional block diagram of the controller as a control means. (A) is a flowchart which shows the procedure which discriminate | determines the internal defect of the fruit in 1st Embodiment, (b) is a flowchart which shows the procedure which discriminate | determines the internal defect of the fruit in 2nd-4th embodiment. It is explanatory drawing of the screen which shows the transmission image information in 1st Embodiment. It is explanatory drawing of the screen which shows smoothed image information, (a) is a case where the smoothing filter of order N = N1 is used, (b) is a case where the smoothing filter of order N = N2 is used, (c) Is a case where a smoothing filter of order N = N3 is used. It is explanatory drawing of the smoothing process with respect to transparent image information. It is explanatory drawing of the screen which shows difference image information, When (a) subtracts the smoothing image information corresponding to FIG.5 (b) from the smoothing image information corresponding to FIG.5 (a), (b) is shown. This is a case where the smoothed image information corresponding to FIG. 5C is subtracted from the smoothed image information corresponding to FIG. It is explanatory drawing of the screen which shows binarized image information. It is explanatory drawing of the screen which shows the image information labeled. It is explanatory drawing of the screen which shows the transmission image information in 2nd Embodiment. It is explanatory drawing of the screen which shows smoothed image information, (a) is a case where the smoothing filter of order N = N1 is used, (b) is a case where the smoothing filter of order N = N2 is used, (c) Is a case where a smoothing filter of order N = N3 is used. It is explanatory drawing of the screen which shows difference image information, When (a) subtracts the smoothing image information corresponding to FIG.11 (b) from the smoothing image information corresponding to FIG.11 (a), (b) is shown. This is a case where the smoothed image information corresponding to FIG. 11C is subtracted from the smoothed image information corresponding to FIG. It is explanatory drawing of the screen which shows binarized image information. It is explanatory drawing of the screen which shows the image information labeled. It is explanatory drawing of the screen which shows the transmission image information in 3rd Embodiment. It is explanatory drawing of the screen which shows smoothed image information, (a) is a case where the smoothing filter of order N = N1 is used, (b) is a case where the smoothing filter of order N = N2 is used, (c) Is a case where a smoothing filter of order N = N3 is used. It is explanatory drawing of the screen which shows difference image information, When (a) subtracts the smoothing image information corresponding to FIG.11 (b) from the smoothing image information corresponding to FIG.11 (a), (b) is shown. This is a case where the smoothed image information corresponding to FIG. 11C is subtracted from the smoothed image information corresponding to FIG. It is explanatory drawing of the screen which shows binarized image information. It is explanatory drawing of the screen which shows the image information labeled. It is explanatory drawing of the screen which shows the labeled image information in 4th Embodiment. It is a flowchart which shows the discrimination | determination procedure in 5th Embodiment. It is explanatory drawing of the screen which shows transparent image information. It is explanatory drawing of the screen which shows binarized image information. It is explanatory drawing of the screen which shows the image information labeled. It is a flowchart which shows the procedure of a judgment process.

A, A 'Area of secondary defect candidate area C Fruit F Set aspect ratio H Set horizontal length M Smoothing filter N Order S Set area L, l Vertical length H, h Horizontal length L / H, l / h Aspect ratio DESCRIPTION OF SYMBOLS 1 Quality inspection apparatus 2 Conveyor line 7 Controller 13 X-ray irradiator 14 X-ray imager 17, 27, 37, 57 Fruit area 18, 28, 38, 58 Line area 19, 29, 39, 59 Defect area 21, 31 , 41 Smoothed fruit region 22, 32, 42 Smoothed stripe region 23, 33, 43 Smoothed defect region 24, 34, 44 Primary defect candidate region 25, 35, 45, 55, 65 Binary fruit region 26 , 36, 46, 56 Secondary defect candidate area 66 Defect candidate area

Claims (2)

An imaging step of irradiating the chestnut fruit with electromagnetic waves and capturing an electromagnetic wave image transmitted through the fruits;
A smoothing step of smoothing the transmission image information obtained by the imaging with a plurality of smoothing filters and specifying a fruit region from each smoothed image information,
Any one of the smoothed fruit regions and the other smoothed fruit regions are subjected to a difference calculation, and each extracted region extracted by the difference calculation is specified as a primary defect candidate region. Differential process;
A binarization process for binarizing the transmission image information and identifying a fruit region from the binarized image information;
An area calculating step for determining an area of each secondary defect candidate area, with each isolated area of the high density pixel surrounded by the low density pixels in the binarized fruit area as a secondary defect candidate area,
If the area of the second defect candidate area corresponding to the first defect candidate area is outside the preset area, the first and second defect candidate areas corresponding to each other are A first determination step for determining that the fruit has an internal defect;
With
Chestnut quality inspection method. About each said secondary defect candidate area | region, the length measurement process which measures the vertical length along the vertical axis direction which goes to the seat from the head in the said fruit, and the horizontal axis direction orthogonal to the said vertical axis direction When,
An aspect ratio calculating step for obtaining an aspect ratio of the length to the width;
The aspect ratio of the second defect candidate area in a fruit other than the fruit determined to be an internal defect in the first determination step is out of a preset set aspect ratio, or the second defect candidate area A second determination step of determining that the second defect candidate area corresponds to an internal defect of the fruit if the horizontal length is larger than a preset set horizontal length;
With
The chestnut quality inspection method according to claim 1.

Images