JP7471216B2 - Crop sorting equipment - Google Patents

Description

The present invention relates to a crop sorting device that evaluates crops such as carrots and radishes using photographed images of the crops being transported and selects the crops based on the evaluation results.

Patent Document 1 discloses an automatic crop sorting device that removes crops one by one from a storage area, weighs and measures them, and sorts and collects them by weight class.

Patent Document 2 discloses an inspection device that uses multiple cameras to capture images of an inspection object transported by a transport means such as a conveyor, and performs an appearance inspection of the fruit or vegetable that is the inspection object based on the multiple captured images. This inspection device includes an input section into which the fruit or vegetable is input, a transport section that transports the input fruit or vegetable, an imaging section that photographs the transported fruit or vegetable, an identification section that uses the multiple captured images of the fruit or vegetable to identify the size, shape, and damage of the fruit or vegetable, and a sorting section that sorts the fruit or vegetable based on the identification results. The size determination by the identification section is performed by comparing the number of pixels in the fruit or vegetable area obtained by image processing of the captured image as the area with a threshold value. The shape determination by the identification section is also performed using the ratio of the long side to the short side of a rectangle that borders the fruit or vegetable area. The damage determination by the identification section is performed by comparing the area of the connected area of the damaged image obtained by image processing with a standard value.

Japanese Patent Application Laid-Open No. 6-156654 Japanese Patent No. 4649628

In the automatic crop sorting device described in Patent Document 1, it is necessary to provide a balance arm for each weight rank for weighing and sorting, so as the number of weight ranks increases, the number of balance arms also increases, and the device becomes larger.

In the inspection device of Patent Document 2, the detected size, shape and damage of fruits and vegetables are compared with standard values to determine the size, shape and damage. Depending on the type and variety of agricultural produce, it may be desirable to sort the produce using the original evaluation criteria of the agricultural producer or manager. To meet such a requirement, a user-friendly user interface for subjectively setting evaluation criteria is required. Such a user interface is not disclosed in Patent Documents 1 and 2.

The object of the present invention is to provide a crop sorting device equipped with a user interface suitable for subjectively setting evaluation criteria.

The crop sorting device of the present invention includes a photographing unit that outputs photographed images of the crops, a crop evaluation unit that outputs evaluation information of the crops based on the photographed images, a reference image management unit that manages reference images for setting evaluation criteria for the crops used in the crop evaluation unit, a display unit that displays the reference images imported from the reference image management unit, and an evaluation criteria setting unit that generates the evaluation criteria based on operational input for the reference image displayed on the display unit and sets the generated evaluation criteria in the crop evaluation unit.

In this configuration, a worker such as a crop producer or crop manager can generate evaluation criteria that are in line with the worker's subjective opinion by performing operational inputs that satisfy the worker's needs with respect to the reference image displayed on the display unit. By setting this evaluation criteria in the crop evaluation unit, the agricultural produce sorting device can sort the crops desired by the worker.

When there are multiple levels of evaluation criteria, it is convenient to prepare and manage in advance many reference images required to generate evaluation criteria for each level from the reference image displayed on the display unit. In addition, since it is the worker who decides the evaluation criteria, it is convenient for the worker to modify the reference image displayed on the display unit himself and create data for determining the evaluation criteria from the modified reference image. By generating evaluation criteria from this data, evaluation criteria that are in line with the worker's subjectivity are generated. For this reason, in a preferred embodiment, the reference image management unit manages the reference image and the approximate outline of the reference image, and the evaluation criteria setting unit generates the evaluation criteria from an adjusted outline obtained by enlarging or reducing the approximate outline displayed on the display unit superimposed on the reference image through an operation input.

For agricultural products such as carrots, it is common to classify the size into three classes. In order to create an adjusted outline suitable for the three-level evaluation criteria from a reference image, it is convenient to create a reduced outline smaller than the reference image and an enlarged outline larger than the reference image, and to create a standard outline between the reduced outline and the enlarged outline. For this reason, in a preferred embodiment, the adjusted outline includes an enlarged outline larger than the reference image, a reduced outline smaller than the reference image, and a standard outline having an area between the enlarged outline and the reduced outline, and the enlarged outline is a large-size evaluation criterion for the area and thickness of the agricultural product, the reduced outline is a small-size evaluation criterion for the area and thickness of the agricultural product, and the standard outline is a medium-size evaluation criterion for the area and thickness of the agricultural product. This configuration has the advantage that when one of the three-level evaluation criteria is to be corrected, it is only necessary to readjust the corresponding outline.

Surface abnormalities of agricultural crops have been visually inspected for a long time, and the evaluation criteria vary depending on conditions such as the type of agricultural crop, its variety, and growing season (early or late). For this reason, surface abnormalities must be evaluated using evaluation criteria that are set in consideration of such conditions. It is preferable that the evaluation criteria be set based on a photographed image of the agricultural crop having a surface abnormality to be judged at the appropriate time. For this reason, in a preferred embodiment, the surface abnormality calculation unit has a comparison calculation function that calculates the surface abnormality by comparing the agricultural crop area in the photographed image with a preselected reference image for abnormality judgment. The reference image for abnormality judgment here may be a photographed image of the agricultural crop that is actually considered to have a surface abnormality, or it may be a pseudo-created image of the agricultural crop that is actually considered to have a surface abnormality.

To easily select an abnormal reference image suitable for determining the evaluation criteria for surface abnormalities from the group of abnormal reference images, it is convenient if the group of abnormal reference images is displayed on the screen of the display unit in sequence according to the level (strength) of abnormality. For this reason, in a preferred embodiment, the display unit displays the abnormal reference images as the group of abnormal reference images in sequence according to the level of abnormality.

FIG. 2 is a schematic perspective view of the crop sorting device. FIG. 2 is a functional block diagram showing a control system of the crop sorting device. FIG. 13 is a schematic diagram showing a screen for setting evaluation criteria for size determination. FIG. 13 is a schematic diagram showing a screen for setting an evaluation criterion for determining a contour shape. FIG. 13 is a schematic diagram showing a sorting destination setting screen. FIG. 4 is a schematic diagram showing a driving status display screen. FIG. 1 is an explanatory diagram for explaining a method for evaluating lumps. FIG. 13 is an explanatory diagram for explaining a method for evaluating a curved root. FIG. 1 is an explanatory diagram for explaining a method for evaluating fork roots.

Below, as an embodiment of the agricultural product sorting device, a carrot sorting device that sorts carrots based on size and surface abnormalities will be described. FIG. 1 is a schematic perspective view of the carrot sorting device. The carrot sorting device is composed of a supply unit 11, an alignment unit 12, a conveying unit 13, and a sorting unit 14. The supply unit 11 is formed as a storage hopper. The alignment unit 12 aligns the carrots conveyed upward from the supply unit 11 and sends them out one by one. The conveying unit 13 conveys the carrots sent out from the supply unit 11 to the sorting unit 14. The sorting unit 14 has eight sorting destinations formed along the conveying direction of the carrots. The sorting unit 14 distributes the conveyed carrots to appropriate sorting destinations based on evaluation information generated by the agricultural product evaluation process described later.

The carrot sorting device is provided with an operation panel 15 to set the speed and other settings of the device. Furthermore, since photographed images of carrots are used to calculate the size grade and surface abnormality level of carrots as evaluation information for carrots, an image capturing unit 2 including a camera for capturing images of the carrots being transported is provided above the transport unit 13.

Figure 2 shows a functional block diagram of the control system of the agricultural product sorting device. This control system includes a sorting machine control device 10. The sorting machine control device 10 receives control parameters from an operation panel 15 and a sorting control unit 100, and controls the carrot sorting device. The sorting control unit 100 acquires photographed images of the carrots from the photographing unit 2, calculates the size grade and surface abnormality state of the carrots, and sends the calculation results to the sorting unit 14 as evaluation information. The sorting control unit 100 includes a agricultural product evaluation unit 3, an evaluation criterion setting unit 4, and a reference image management unit 6. It also includes a touch panel unit 5 that functions as a display unit that allows the operator to input information and displays information to the operator.

The crop evaluation unit 3 has a size calculation unit 31 and a surface abnormality calculation unit 32. The evaluation criteria used by the size calculation unit 31 and the surface abnormality calculation unit 32 are set by the evaluation criteria setting unit 4. The captured image of the carrot from the photographing unit 2 is provided to the size calculation unit 31 and the surface abnormality calculation unit 32. The size calculation unit 31 calculates the size grade of the carrot using the evaluation criteria and the captured image, and the surface abnormality calculation unit 32 calculates the surface abnormality of the carrot using the evaluation criteria and the captured image. The crop evaluation unit 3 generates evaluation information as an evaluation result of the carrot based on the calculation results of the size calculation unit 31 and the surface abnormality calculation unit 32, and outputs it to the sorting unit 14.

The size calculation unit 31 calculates a size grade based on the outline of the carrot calculated from the photographed image of the carrot and the area of the internal region of this outline. The size grades are "SSS", "SS", "S", "M", "L", "LL", and "LLL", in order from smallest to largest. A grade selected from these is actually assigned to the sorting destination by the sorting unit 14. Note that a carrot that is much longer and thinner than a standard carrot is not necessarily considered to be of the same size grade even if it has the same area as a standard carrot. Also, depending on the type of vegetable or fruit, the size of some is evaluated by weight rather than by apparent size. For this reason, the size calculation unit 31 also has a function as an option to estimate and calculate the weight based on the outline calculated from the photographed image and the area within this outline, and to calculate the size grade from the estimated calculation result.

In this embodiment, the surface abnormality calculation unit 32 has three main image processing functions to calculate the surface abnormality of the carrot. The first function is a function to calculate a specific surface abnormality based on the shape characteristics of the outline of a specific part obtained by dividing the outline of the carrot region extracted from the photographed image into multiple parts. The second function is a function to calculate a specific surface abnormality based on the area and number of image singular points in the carrot region extracted from the photographed image. The third function is a function to calculate a specific surface abnormality based on the distribution degree of a specific color in a specific part of the carrot region extracted from the photographed image.

The size calculation unit 31 and surface abnormality calculation unit 32 that make up the agricultural crop evaluation unit 3 evaluate the carrots based on the evaluation criteria set by the evaluation criteria setting unit 4. The evaluation criteria are generated by the evaluation criteria setting unit 4 based on instruction information from the touch panel unit 5. The touch panel unit 5 generates instruction information based on operational input by the worker and provides it to the evaluation criteria setting unit 4.

The touch panel unit 5 has a display control unit 51, a flat panel display 50 abbreviated as FPD, and an operation unit 52. The flat panel display 50 functions as a touch panel. The operation unit 52 has software switches displayed on the flat panel display 50 and hardware switches provided around the flat panel display 50. The display control unit 51 generates various information to be displayed on the flat panel display 50. Furthermore, the display control unit 51 outputs processed information obtained by processing operation input by the operation unit 52 to other functional units. In other words, the touch panel unit 5 forms a graphic user interface for the sorting control unit 100.

The reference image management unit 6 manages the reference images described below. The reference images captured from the reference image management unit 6 are displayed on the flat panel display 50. One example of a reference image is an actual reference image that can be visually identified as a carrot, and is a photographed image showing various sizes of carrots or an approximate outline of a reference image representing a carrot. Instead of a photographed image, an illustrated image may be used as the actual reference image. Another example of a reference image is a photographed image of a carrot showing various surface abnormalities (an abnormal reference image). For this abnormal reference image, an illustrated image may also be used instead of a photographed image.

Next, we will explain the evaluation criteria generation process, which generates evaluation criteria through cooperation between the reference image management unit 6, the touch panel unit 5, and the evaluation criteria setting unit 4.

(Evaluation criteria for size evaluation)
In the process of generating an evaluation standard for size evaluation, a screen of a flat panel display 50 as shown in FIG. 3 is used. In the center of the screen, an actual reference image showing a carrot of a standard size is displayed, and a plurality of (four in FIG. 3) approximate outlines of the carrot that are simplified in outline are displayed superimposed on the actual reference image. These approximate outlines are also managed by the reference image management unit 6. In addition, in the lower region of the screen, a first horizontal bar for setting the grade of the area (projected area) of the carrot and a second horizontal bar for setting the grade of the thickness (width) of the carrot are displayed. A plurality of (three in FIG. 3) area cursors are arranged on the first horizontal bar so as to be movable along the first horizontal bar by the operator's operation. A plurality of (three in FIG. 3) thickness cursors are also arranged on the second horizontal bar so as to be movable along the second horizontal bar by the operator's operation. The plurality of area cursors are for setting the area grades. The position of each area cursor on the first horizontal bar is linked to the area of the approximate outline assigned to each area cursor, and the area grades such as S, M, L, and LL are classified according to the position of each area cursor. Similarly, the thickness cursors set the thickness grade. The position of each thickness cursor on the second horizontal bar is linked to the width of the approximate outline assigned to each area cursor, and the thickness grades such as S, M, L, and LL are classified according to the position of each thickness cursor. Depending on the position of this thickness cursor, the size grade can be limited by the thickness, rather than setting the size grade only by the area. This makes it possible to lower the grade of a carrot that is too thin even if it has the same area.

Each approximate outline is enlarged or reduced by operating (operation input) the area cursor and thickness cursor. The final adjusted approximate outlines, which are the multiple adjusted outlines, are sent to the evaluation criterion setting unit 4. The evaluation criterion setting unit 4 processes data related to the adjusted outlines as instruction information, generates an evaluation criterion related to size, and sends it to the agricultural crop evaluation unit 3. This evaluation criterion related to size is set in the size calculation unit 31. The size calculation unit 31 compares the image processing results of the captured image sent from the photography unit 2 with the evaluation criterion, and calculates the size grade of the carrots being transported.

Specifically, the adjusted outline includes one or more enlarged outlines larger than an actual reference image (an example of a reference image) showing a carrot of a standard size, one or more reduced outlines smaller than the actual reference image, and a standard outline having an area between the enlarged outline and the reduced outline. The enlarged outline serves as an evaluation criterion for large sizes (LL or L) in terms of the area and thickness of the carrot, the reduced outline serves as an evaluation criterion for small sizes (S or SS) in terms of the area and thickness of the carrot, and the standard outline serves as an evaluation criterion for medium sizes (M) in terms of the area and thickness of the carrot. Note that instead of various outlines such as the reduced outline, standard outline, and enlarged outline, data of a reference image corresponding to each outline may be sent to the evaluation criterion setting unit 4 as instruction information. In this case, the evaluation criterion setting unit 4 generates an evaluation criterion based on the data of the reference image sent as instruction information and sets it in the agricultural product evaluation unit 3.

(Evaluation criteria for evaluating surface abnormalities)
In the process of generating an evaluation standard for evaluating surface abnormalities, for example, a screen of a flat panel display 50 as shown in FIG. 4 is used. In the example shown in FIG. 4, a group of abnormal reference images (including non-abnormal reference images for comparison) showing various forms of carrots are arranged in a matrix (5 rows x 6 columns in FIG. 4), and a horizontal judgment bar is arranged in the area below the group. A judgment cursor is arranged on the horizontal judgment bar so as to be movable along the horizontal judgment bar by an operator's operation. The judgment cursor can be moved to the boundary position of each column of carrots arranged in the matrix. This screen is used to judge the state of a bump as a surface abnormality of a carrot. Therefore, in the central area, carrots with bump states ranging from normal to abnormal are arranged in stages. In the rightmost column in the central area, carrots considered to be the most normal grade (OK) are arranged, and carrots considered to be the most abnormal grade (NG) are arranged in the leftmost column. In each column, carrots of the same grade but different normal or abnormal states are arranged. That is, they are displayed in order according to the magnitude of the abnormality. This allows the operator to easily visually recognize the level of abnormality of various types of carrots. The abnormal reference images are managed in the reference image management unit 6 with attribute values that indicate the level of abnormality.

The operator moves the judgment cursor to a position that is the boundary between OK and NG. Based on the final position of the judgment cursor, the touch panel unit 5 generates instruction information for evaluating the lumps and sends it to the evaluation criterion setting unit 4. Based on the instruction information, the evaluation criterion setting unit 4 generates evaluation criteria for lumps, which are an example of a surface abnormality, and sends it to the agricultural product evaluation unit 3. This evaluation criterion for lumps is set in the surface abnormality calculation unit 32. The surface abnormality calculation unit 32 processes the captured image sent from the imaging unit 2, and compares the image processing results with the set evaluation criteria to calculate the surface abnormality state of the photographed carrot.

Surface abnormalities on carrots include dents and curved roots in addition to lumps, but even if these are used as selection criteria, the operator can generate and set the desired evaluation criteria using the abnormal reference images and judgment cursor, as described above.

Based on the size class calculated by the size calculation unit 31 and the surface abnormality state calculated by the surface abnormality calculation unit 32, evaluation information specifying the sorting destination is created and provided to the sorting unit 14.

To set the sorting destinations in the sorting unit 14, for example, a sorting destination setting screen as shown in FIG. 4 is used. In this example, there are eight sorting destinations, and the sorting conditions are size class and type of surface abnormality. Size classes ranging from SSS to LLL can be selected as size classes, and surface abnormality types such as no abnormality, curved root, forked root, stain, scratch, poor coloring, and blue neck can be selected. For example, depending on the settings, carrots with size class M and no surface abnormalities are sorted to sorting destination "1," and carrots with size class L and abnormal lumps are sorted to sorting destination "5."

The operating status of the carrot sorting device can be confirmed and changed through the operating screen shown in FIG. 6. In this embodiment, the central area of the operating screen displays the captured image from the image capture unit 2. The lower area displays a group of buttons for adjusting the processing speed of the device. The group of buttons includes a stop button, a pause button, a start button, a speed down button, and a speed up button.

(Image Processing Algorithm)
An example of an image processing algorithm implemented in the crop assessment unit 3 for detecting surface anomalies will now be described.
(1) Bumps and dents (peripheral area)
In detecting lumps and dents as surface abnormalities of a carrot as shown in FIG. 7, the tip and root regions of the carrot are excluded. The lengths of the excluded tip and root regions are each p% of the total length. This p can be set arbitrarily by the operator. The remaining intermediate region excluding the tip and root regions is the detection target region. In an actual photographed image, the tip and root regions may be reversed, so the tip and root regions do not need to be distinguished. Alternatively, since a carrot is cone-shaped, the image may be rotated so that the tip region is automatically on one side. The evaluation of the lumps in the outer peripheral region shown in FIG. 7 is performed by fragmenting the inner region of the contour of the photographed image of the carrot, which corresponds to the cross section of the carrot, in the transverse direction with a width of a predetermined number of pixels, and based on the length from the virtual center line P1 of the carrot to the apex of each fragment. Specifically, the maximum value in the envelope of the apex of each fragment within a predetermined range is found. Furthermore, a judgment circle of a predetermined radius (indicated by r in FIG. 7) is set with the apex of the fragment having this maximum value as its center. The ratio of the carrot area to the background area in this judgment circle is calculated, and if the ratio of the carrot area is equal to or less than the bump threshold, it is judged to be a bump. If the number of bumps extracted in this way exceeds the bump judgment value, the carrot is judged to have a surface abnormality (bump) and is selected. Note that the evaluation of dents is similar to that of bumps, and instead of the maximum value in the envelope of the vertices of each fragment within a specified range, the minimum value is calculated, and the dents are evaluated in the same manner.

(2) Bumps and dents (inner area)
To detect bumps and dents in the internal region of a carrot (the internal region in a photographed image of a carrot), edge detection is used, utilizing the brightness change that occurs at the boundary line of the bump or dent. If the area of the edge-detected region is equal to or greater than a predetermined value, it is deemed to be a bump or dent, and if the number of bumps or dents exceeds a bump (dent) judgment value, the carrot in the photographed image is evaluated as having a surface abnormality (bump, dent) and is selected out. Note that blemishes on the surface of the carrot are also evaluated in a similar manner.

(3) Curved Roots In detecting curved roots as a surface abnormality of a carrot as shown in FIG. 8, the tip and root regions of the carrot are also excluded. An approximation line of the outer contour of the middle region, which is the region to be evaluated, is calculated. Here, the approximation line is a line connecting the boundary point between the root region and the middle region and the boundary point between the middle region and the tip region. Next, the deviation of the extracted outer contour from the approximation line is calculated. In calculating this deviation, the length of each fragment (the length from the virtual center line P1 to the apex) obtained by fragmenting the inner region of the contour of the photographed image of the carrot with a width of a predetermined number of pixels is used. The area between the approximation line and the apex of the fragment (the bulging portion is classified as a positive area and the depressed portion is classified as a negative area) is used as the deviation area and is integrated over all the fragments. Note that the length of the fragments in the region close to the root region or the tip region contains a large error, so the averaged length is used. If the integrated deviation area exceeds the curved root judgment value, the carrot is evaluated as having a surface abnormality (curved root) and is selected.

(4) Forks As shown in Fig. 9, forks are detected as surface abnormalities of carrots by setting transverse lines (pixel lines shown by dotted lines in Fig. 9) at intervals of a predetermined number of pixels along the longitudinal direction of the carrot, and if a carrot part and a background part are repeated on the transverse line, the transverse line is regarded as a fork crossing line. If the number of these fork crossing lines exceeds the fork judgment value, the carrot is evaluated as having a surface abnormality (forks) and is selected. The number of these fork crossing lines represents the length (number of lines x predetermined number of pixels), so in Fig. 8, the fork judgment value is represented by the length q, and the actually calculated length (number of fork crossing lines) is represented by r.

(5) Scratches Edge detection is used to detect forks as surface abnormalities of carrots, in the same way as detecting lumps and dents in internal regions. Closed regions obtained by edge detection are regarded as scratch regions, and the lengths and number of scratch regions are calculated. If the number of scratch regions having lengths exceeding the first scratch judgment value exceeds the second scratch judgment value, the carrot is evaluated as having a surface abnormality (scratches) and is selected.

(6) Coloring Defects As with the detection of blemishes and scratches, edge detection is also used to detect coloring defects as a surface abnormality of carrots. Threshold values for hue, saturation, and brightness that indicate coloring defects are set as edge thresholds for edge detection. If the area of a coloring defect region obtained by edge detection exceeds the coloring defect judgment value, the carrot is evaluated as having a surface abnormality (coloring defect) and is selected.

(7) Green Neck When detecting a green neck as a surface abnormality of a carrot, the carrot is divided into a tip region, a middle region, and a base region. If the tip region and the base region are not distinguished in the captured image, both the tip region and the base region become the detection target region, and if the tip region and the base region are distinguished, the base region becomes the detection target region. In the detection target region, the captured image is ternarized with the background color, a bluish color, and an orange color. If the proportion of blue-colored pixels in the ternarized image data exceeds the blue neck judgment value, the carrot is evaluated as having a surface abnormality (green neck) and is selected.

In the image processing algorithm described above, various judgment values for detecting surface abnormalities may be set as default values. Alternatively, an operator may select the boundaries of surface abnormalities from a group of abnormal reference images in which surface abnormalities are arranged by abnormality level as shown in FIG. 5, and the surface abnormality judgment values and various thresholds may be set based on the selection result. For example, in the case of detecting coloring defects, data for the threshold value of coloring defects is created from a photographed image that is considered to have a coloring defect and is selected by the operator from a group of photographed images showing actual coloring defects in stages as shown in FIG. 5.

[Another embodiment]
(1) In the above-described embodiment, carrots are used as the agricultural product to be sorted. However, other fruits and vegetables may also be used.
(2) The various functional blocks shown in FIG. 2 are for illustrative purposes only, and these functional blocks may be combined with other functional blocks or divided into multiple sub-blocks.

The configurations disclosed in the above embodiment (including other embodiments, the same applies below) can be applied in combination with configurations disclosed in other embodiments, so long as no contradiction arises. Furthermore, the embodiments disclosed in this specification are merely examples, and the present invention is not limited to these embodiments. They can be modified as appropriate within the scope of the purpose of the present invention.

The present invention can be applied to a crop sorting device that uses photographed images of crops to evaluate the crops and sort them based on the evaluation results.

2: Photographing section 3: Crop evaluation unit 4: Evaluation standard setting section 5: Touch panel unit 6: Reference image management section 10: Sorting machine control device 11: Supply section 12: Alignment section 13: Transport section 14: Sorting section 31: Size calculation section 32: Surface abnormality calculation section 50: Flat panel display 51: Display control section 52: Operation section 100: Sorting control unit

Claims (5)

an image capturing unit that outputs captured images of agricultural crops;
a crop evaluation unit that outputs evaluation information of the crop based on the captured image;
a reference image management unit that manages reference images for setting evaluation criteria for the agricultural crops used in the agricultural crop evaluation unit;
a display unit that displays the reference image acquired from the reference image management unit;
an evaluation criterion setting unit that generates the evaluation criterion based on an operation input for the reference image displayed on the display unit and sets the generated evaluation criterion in the agricultural product evaluation unit;
Equipped with
the reference image management unit manages the reference image and an approximate outline of the reference image;
The evaluation criterion setting unit generates the evaluation criterion from an adjusted outline obtained based on the approximate outline displayed on the display unit so as to be superimposed on the reference image . The farm product selection device according to claim 1 , wherein the evaluation criterion setting unit generates the evaluation criterion from an adjusted outline obtained by enlarging or reducing the approximate outline through an operational input. The adjusted outline includes an enlarged outline larger than the reference image, a reduced outline smaller than the reference image, and a standard outline having an area between the enlarged outline and the reduced outline, the enlarged outline serving as a large-size evaluation standard for the area and thickness of the crop, the reduced outline serving as a small-size evaluation standard for the area and thickness of the crop, and the standard outline serving as a medium-size evaluation standard for the area and thickness of the crop. an image capturing unit that outputs captured images of agricultural crops;
a crop evaluation unit that outputs evaluation information of the crop based on the captured image;
a reference image management unit that manages reference images for setting evaluation criteria for the agricultural crops used in the agricultural crop evaluation unit;
a display unit that displays the reference image acquired from the reference image management unit;
an evaluation criterion setting unit that generates the evaluation criterion based on an operation input for the reference image displayed on the display unit and sets the generated evaluation criterion in the agricultural product evaluation unit;
Equipped with
the reference image management unit manages, as the reference images, a plurality of abnormal reference images each showing a stage of surface abnormality of the crop;
the display unit displays an abnormal reference image group consisting of a plurality of the abnormal reference images;
The evaluation criterion setting unit sets the evaluation criterion from the reference image selected by operational input from the group of abnormal reference images . The crop sorting device according to claim 4, wherein the display unit displays the abnormal reference images as the abnormal reference image group in order according to the magnitude of the degree of abnormality.

Images