JPS62250981A - Selector - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a sorting device that can automatically sort objects such as grapes in which a plurality of objects to be inspected overlap each other.

[Background of the Invention] Currently, in grape sorting plants, inspectors generally perform fruit sorting and evaluation visually. That is, the inspector evaluates the color of the grapes using a color sample, and then sorts the grapes into each grade based on evaluations of the size of the grapes, the shape of the bunch, etc.

[Problems to be Solved by the Invention] By the way, during the peak grape season, approximately 10,000 boxes of grapes must be inspected per day, so the conventional sorting method described above does not allow for the physical condition of the inspectors such as fatigue. Inspection variations occur due to individual differences among inspectors.This invention was made against this background. It is an object of the present invention to provide a sorting device that can perform automatic sorting without variations in evaluation results.

[Means for Solving the Problems] In order to solve the above-mentioned problems, the present invention provides an imaging means for imaging a plurality of objects to be inspected that are positioned one on top of the other like the fruits of a grape; extracting means for extracting the outline of each of the objects to be inspected; shape determining means for determining the size of the object to be inspected and its variation from the outline; comprising a color difference meter that obtains a signal corresponding to the color tone of the body; and a color tone determination means for determining the color tone of the object to be inspected based on the output of the color difference meter, and based on the outputs of the shape determination means and the color tone determination means. The gist is to select the objects to be inspected.

Further, it has an object determining means for extracting an object mixed in the object to be inspected, such as a stem of a bunch of grapes, from the output of the imaging means and outputting a mixing ratio of the extracted object and the object to be inspected, The object to be inspected is selected based on the outputs of the shape determining means, color tone determining means, and object determining means.

[Function] According to the above configuration, the quality of the object to be inspected, such as grapes, can be determined objectively and automatically by comparing the external shape and color tone of the object to be inspected with predetermined evaluation criteria. Can be done. In particular, even when the objects to be inspected overlap each other, the contours of each object to be inspected are extracted and judged, making it possible to make accurate judgments.

In addition, the quality of the inspected object is determined by determining the mixing ratio of the object mixed with the object being inspected, such as the stem of a bunch of grapes, and using this mixing ratio in conjunction with the shape and color mentioned above. , enabling more accurate quality determination.

[Example] Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a plan view showing the overall configuration of a sorting device according to an embodiment of the present invention. In the figure! is a conveyor bent in an L shape. The conveyor 1 sequentially transports grapes that are housed in cardboard boxes and carried in.

Various devices are disposed along the path of the conveyor I. First, a barcode reader 2 is provided to read the barcode attached to the cardboard box, and the manufacturer code and box type (1, 2, 4, 8 kg) are read. Further, a darkroom 3 is provided downstream of the barcode reader 2, and a CCD camera 4 for taking images of grapes and a color difference meter 5 are installed in this darkroom 3. Note that these details will be described later.

Further downstream of the conveyor 1, a stamp device 6 and a sorting mechanism 7 are installed. The stamp device 6 is used to stamp a grade indication on the cardboard box, and the sorting mechanism 7 is used to sort grapes into first-class, second-class, other, etc. depending on the quality. In addition, prior to this sorting, the 4 kg boxes and other boxes are separated by a preset switch.
The boxes are automatically faced by the opposing machine 8a and then subjected to sorting, while the other boxes pass through a separate route 8b and undergo the above sorting. The sorted 4 kg boxes are packed by a packing machine 9 and shipped. Other boxes are manually packed and shipped. In addition, in the figure, 10 is a computer-controlled device distribution board, +1.12°13.14.1
5 is a worker carrying in cardboard boxes, a worker opening the lid, a worker closing the lid, a worker packing boxes other than 4kg, and a worker carrying 4kg.
He is a worker shipping boxes.

FIG. 2 is a block diagram showing the overall configuration of a control device that controls each part of FIG. 1. In the figure, 20 is an image processing device, 30 is an input/output device control device, and 40 is a management computer.

The image processing device 20 converts the output signal of the CCD camera 4 into red,
Converter 4a that separates green and blue three color signals R, G, and B
It converts and stores the output signals of , and performs certain processing on them. That is, the image quality processing device 20 performs the above n
, G, and A/D converters 21R, 21G, and 21 that convert the I3 signal into a digital signal and output it as image data.
13, and a frame memory 22 that stores this image data.
11, 22G and 22B, a CPU 23 that performs processing to be described later on image data, a ROM 24 that stores basic data necessary for programs and processing of the CPU 23, and a RAM 25 that is a work area for the CPU 23 and the like.

Next, the input/output device control device 30 controls each input/output device arranged along the conveyor 1. That is, a barcode league interface 31 that takes in a signal from the barcode reader 2, converts it into digital data, and stores it, and a color difference meter interface that takes in an analog signal output from the color difference meter 5, converts it into a digital signal, and stores it. 32, a management computer interface 33 that controls the exchange of signals with the management computer 40, an image processing device interface 34 that controls the exchange of signals with the image processing device 20, a line control interface 35 that performs line control, and It is composed of a manual contact group 36 connected to the line control interface 35, an output contact group 37, and a power supply section 38 that supplies power to each of the above sections.

Each of the above-mentioned interfaces 31 to 35 includes a CPU with a suffix a attached to each reference numeral, and a ROM with a suffix suffix attached to each reference numeral.
and RAM with the subscript C. For example, the barcode league interface 31
It is composed of U31a, ROM31b, and RAM31c.

Here, the CPUs of these interfaces 31 to 35
31a to 35a are system buses 39. The CPU 31a,
33a and 34a are a barcode reader 2, a management computer 40, an image processing CPU 23, and an R5-23, respectively.
Communication is performed using two standards.

Further, the CPU 32a has an A/D conversion function of converting an analog signal from the color difference meter 5 into a digital signal, and a function of taking an average of N digital signals. Furthermore, the CPU 35a has a function of processing signals sent through the input contact group 36 from the proximity switches arranged at various locations on the line, and then forming a line control signal and outputting it through the output contact group 37. have. The output contact group 37 consists of switching elements such as relays, is connected to the CPU 35a via a sequence bus 39b, and is connected to the conveyor! , CCD camera 4, color difference meter 5, and sorting mechanism 7. Note that the CCD camera 4 and the color difference meter 5 are driven by a pulse command.
x, y by axis robot.

It can be moved in each Z direction.

The management computer 40 determines the grade of grapes based on the processing results from the image processing device 20 and the color difference meter CPU 32a. In addition, this decision result is applied to the communication CP
It is sent to the line control CPU 35a via U33a, and instructs the sorting mechanism 7 to sort. Furthermore, the number of shipments for each grade for each producer is totaled, and a shipment performance table is created.

FIG. 3 is a memory map showing the configuration of memory within the image processing device 20. As shown in FIG. In the figure, a DRAM area 25a is an area for system programs. Next to this DRAM area 25a is a RAM area 125 for processing programs.
b, RAM area ■25c, and frame memory 22R
, 22G, and 22B are taken. Furthermore, in the next area, place a free area and use RO to start up the system.
An M area 25d and an I10 area 25e for f10 control are secured.

FIG. 4 is a block diagram showing the configuration of the color difference meter 5. In the figure, 51 is a spectroscopic section. Spectral r′! l<51 means that the light emitting element 52 emits white light, the lens 53 that collects the light emitted from the light emitting element 52, and the three filters 54.
B, 54 fl, and 54 Y, the light emitted from the lens 53 is split into three colors of blue, red, and yellow.
It rotates at 0 rpm). Therefore, the monochromatic lights of blue, red, and yellow that have passed through the rotating disk 54 are sequentially irradiated onto the grapes 55 at a constant period. Note that a start mark 54a is provided at a predetermined position on the rotating disk 54.

The light reflected by the grapes 55 is received by the light receiving element 56,
It is converted into an electrical signal and supplied to the amplifier 57. The amplifier 57 amplifies the output voltage of the light receiving element 56 and sends it to the CPU 32a.
(see also Figure 2).

The CPU 32a converts this into 12-bit digital data by A/D conversion, and performs processing to be described later to convert the color of the grapes 55 into numerical values. Note that the light receiving element 56 is an element having flat characteristics for each color.

FIG. 5 shows the output of the color difference meter 5, and the figure (a
) indicates a start signal by the start mark 54a,
FIG. 5B shows the color signal output from the amplifier 57.

As you can see from this diagram, after the start signal, the blue B
, red R1 yellow Y are sequentially output, and these are supplied to the CPU 32a. In addition, start mark 54
a is detected by a light receiving element (not shown), and the CPU 32
It is designed to be supplied to a.

Next, the operation of this embodiment will be explained.

FIG. 6 shows the overall operation of this embodiment, roughly divided into three types: one mainly for human work, one mainly for image processing and line control, and one mainly for the processing of the management computer 40. It is a flowchart.

First, the cardboard box containing the grapes becomes a work opening! When carried in by conveyor l (step St), this conveyor l
and passes through the barcode reader 2. At this time, the barcode written on the cardboard box is read by the barcode reader 2 (step S2)
. This barcode includes information such as the fruit sorting facility code, producer code, box type, and grape bunch size, and the read barcode is sent from the barcode reader 2 to the CPU 31a in FIG. , are sent to the management computer 40 via the CPU 33a. In step S, the management computer 40 determines whether or not the reading has been performed normally.
3) If normal, proceed to step S4; if not, send a line stop command to the line control CPU 35a via the CPU 33a, and the CPU 35a stops the conveyor I (step S7). When the conveyor l is stopped, manual recovery processing and reinjection are performed (step S8).
~9), conveyor l is restarted.

In step S4, the management computer 40 searches the master file in which the producer is registered, checks whether the fruit sorting facility code, producer code, box type, etc. If so (step S5), the fruit sorting place, producer code, box type, and bunch size are recorded in step S6, and the process enters a waiting state.

At this time, control proceeds to step Si1. On the other hand, if the barcode is not genuine, the management computer 4
0 through the CPU 33a, the line stop command is sent to the CPU 33a.
35a, and the CPU 35a stops the conveyor I (step S7). When the conveyor 1 is stopped, recovery processing and reinjection are performed manually (steps 88 to 8).
9), conveyor! will be restarted.

Now, in step Sll, the operator 12 opens the lid. Thereafter, the COD camera 4 images the grapes, and this image data is transferred to the image processing device 20. Based on this image data, the image processing device 20 measures the diameter of the grape grain (step 512) and measures the stem exposure ratio (step S13). Next, the color tone of the grapes is detected by the color difference meter 5, and this detection result is converted into digital data by the CPU 32a.
Color tone measurement is performed based on the digital data (
Step 514). Note that these processes will be detailed later.

The above processing result is the CI of the image processing device 20) U 23
and the management computer 4 from the color tone measurement CPU 32a.
0 via the communication CPO 33a. The management computer 40 compares the above processing results with the data in the master file to determine the grade (step 515), and records it as performance data for each producer (step S16).

When the performance data is obtained, the management computer 40 sends an instruction to handle the line to the line control CPU 35a through the communication CPU 33a (step S17). CP for line control
U35a operates the stamp device 6 to stamp the grade stamp on the corresponding cardboard box (step 518).

After the grade stamp is stamped, the operator 13 closes the lid (step 521). Also, a distinction is made between 4 kg boxes and other boxes (1 kg, 2 kg, 8 kg boxes) (step 522), and after the 4 kg boxes are automatically imaged by the opposing machine 8a (step 523), other boxes are Another route 8
b, the fruit reaches the sorting mechanism 7, where it is sorted (step 524). Then, it is packed in step S25 and shipped in step S26.

Note that the above-mentioned 4 kg boxes are distinguished from other boxes by a manually set switch before driving the conveyor I. For example, if the weight is set to 4 kg, only 4 kg boxes are carried in.

Next, FIG. 7 is a flowchart showing an overview of image processing by the image processing device 20, and this includes steps S12 to ! of FIG. 6. This corresponds to 4.

In this case, when the cardboard box approaches the COD camera 4,
This is detected by the proximity sensor, and the input contact 36 in Fig. 2 → line control CPU 35a → communication CPU 34a
→A measurement command is output through the image processing CPU 23. When this command is detected (step 5I131),
The CPU 23 performs pre-processing (step 5B2), calculates the diameter of the grape fruit (step 5n3), calculates the exposure ratio of the stem of the grape bunch (step 5B4), and sends the calculation results to the management computer 40 (step 5B5). .

Each of the above processes will be explained below.

FIG. 8 and FIG. 9 are diagrams for explaining the above preprocessing. This preprocessing is to extract the outline of each fruit from the image of the bunch of grapes captured by the COD camera 4. In this case, the outline of the fruit is close to a perfect circle and is much brighter than its surroundings. The number of pixels of the COD camera 4 is 512 x 512, and the brightness of each pixel is θ~255.
That is, it is expressed in 8 bits. Further, the captured grape images are stored in each frame memory 22R to 22B for each color.

When the preprocessing is started, the image processing CPU 23 performs the following processing. First, in step SPI of FIG.
For each pixel of frame memories 22R and 22B,
The logical sum is taken and the result is transferred to the RAM area 125b. Then, as shown in FIG. 9, starting from the second pixel C on the second horizontal line, arithmetic processing is sequentially performed on four adjacent pixels A, B, D, and E. This process involves the process l (step S) of binarizing each pixel and extracting bright pixels as outlines.
Steps SP2 to SP8) and Processing 2 (Steps SP9 to SP16) for making the extracted outline more clear.

In process 1, as shown in FIG. 9(a), the RAM area [
Read out each pixel from 25b and perform the following calculation,
The data is written to the corresponding address in the RAM area 1125c.

First, in step SP2, the luminance level of pixel C is multiplied by 4, and this value is set as X. From this value X, the adjacent pixel A
, B, D, and E to obtain a new value X (step 5P3). If this new value
+ hexadecimal representation) is written to the corresponding address of the RAM area [125c (step SP4.5). On the other hand, when the new value X is smaller than the binarization level, the RAM area
"00" is written to the corresponding address of 25c (step SP4.6). That is, when the luminance level of pixel C is higher than the average of surrounding pixels A to E, "FF" is written, and when it is lower, "00" is written to the corresponding address in the RAM area 1125c.

Thereafter, each pixel is extracted one by one (step 5P7), the above process is repeated, and when this binarization process is completed for all pixels (step 5P8), the process proceeds to the next process 2.

After bright parts are extracted in process 1, the next process is to remove isolated points. This process is performed as shown in FIG.
This is a process of reading out one pixel from the AM area 1125c, performing the following calculations, and writing to the rtAM area I25b.

In this case, the following three types of processing are performed depending on the states of the four adjacent pixels A 1 , B 2 , D 2 , and E.

(1) When pixels A, B, D, and E are all "00", that is, when it is dark, pixel C is also set to "00". This removes isolated points.

(2) When pixels A, B, D, and E are all "FF", that is, high brightness, pixel C is set to "00". This accomplishes thinning.

(3) In cases other than (1) and (2) above, the value of pixel C itself is transferred.

Next, the specific operation of process 2 will be explained with reference to FIG.

First, when pixels A, B, D, and E are all "00", pixels A, B, D, and E are calculated in step SP9.
Since the logical sum X becomes "00", the process moves from step 5PIO to step 5P14, and regardless of the value of pixel C, the above logical sum x1, that is, "00" is stored in the RAM area 125.
transferred to b. This means that the process (1) above has been executed, and isolated high-brightness points are removed.

Next, when pixels A, B, D, and E are all "FF", the logical sum X becomes "FF", so step 5PI
Proceed to step 5PII from O, pixels A, B, D
.

The AND of E is taken. This logical product is also “FF”,
The negative logic “00” is set as a new X (step 5
P12). Therefore, the logical product of the pixel C calculated in step SP+3 and the new value X becomes "00", which is R
The data is transferred to the AM area 125b.

In this way, the process (2) above is performed.

Furthermore, in cases other than the above, that is, the adjacent 4 pixels A.

When one to three of B, D, and E are "FF" (high brightness), there are at least a few "00" pixels, so the logic of four pixels A, B, D, and E is The product is “00”
Then, the negative logic X becomes “FF” (step S
P + 2°13). Therefore, the value of pixel C remains R
Transferred to the AM area 125b (step S13.14
).

Thereafter, the next pixel is sequentially extracted (step SP+5), and the same process as above is performed, and when all pixels have been processed, the preprocessing is completed (step SP I G).

Once the preprocessing is complete, the next step is to calculate the diameter of the grape fruit. The diameter calculation process will be described below with reference to FIGS. 10 to 12.

This processing is performed based on image data captured by the COD camera 4. That is, a circular arc is found by tracing consecutive pixels from the image data, and the diameter of the grape grain is determined from three points on this circular arc. Note that in this case, errors occur because the grape grains are analyzed with perfect circular binding. Taking this error into account, arcs whose center coordinates fall within a certain small range are considered to be the same circle, and the average diameter is taken as the final diameter.

In FIG. 10, pixels are read out in order from the image data stored in the RAM area 125b, and these are read out from the pixel of interest P.
(Step 5P31). This pixel of interest P is “00”
If the pixel of interest is “FF”, the next pixel is read out and set as the pixel of interest P. If the pixel of interest is “FF”, the adjacent pixel is
The following processing is performed for each of pixels A to I-1.

First, select the adjacent pixel A and read it out Step 5P3
3), check whether this is “FF” (step 5P3
4). Then, in the case of "FF", addition of vector quantities is executed (step 5P35). This vector quantity is the first
As shown in FIG. 1(b), the pixel of interest P and the adjacent pixel A −
It is determined by the positional relationship with H. For example, the vector amount of a pixel person is (1, 1), and the vector amount of a pixel F is (1,
-1), the pixel P of interest is determined corresponding to the origin and the LY coordinate. In the current case where pixel A is selected, the vector quantities are VX=1 and Vy=I, so the sum of the vector quantities TVX and TVy can be found by the following equation.

TVx=TVxprev+I--(1)TVy=
T V yprev+ 1- (2) Here,
T V xprev and T V yprev are vector sums up to the number of times each song is played, and the vector sums over the past several times are taken.

Furthermore, when other pixels B to H are selected, their respective vectors ff1Vx and Vy are added in the same manner to obtain vector sums TVX and TVy.

Based on the vector sums TVX and TVY thus obtained, the analysis direction is determined as shown in FIG. 11(c) (step 5P36). In this figure, the vector sum T
Depending on the sign (positive or negative) of Vx and TVy and the ratio, -period is 2
It is divided into 16 directions at 2.5 degrees, and the next pixel to be selected is determined. For example, vector sum TVx, TVY
are both positive, and their ratio TVy/TVxh(tan
22, if it is smaller than 5°, step SP37-3
9, the process returns to step 5P33, and pixels A, F' are selected one after another in this order, and continuous pixels are tracked.

In this way, once several tens of continuous pixels have been tracked (step 5P37), this continuous pixel is regarded as an arc, and the diameter and center coordinates are calculated by the method described later (step 5P37).
P38). When this process is completed for the above-mentioned 8 adjacent pixels A to ■] (step 5P39), the next pixel of interest is sequentially selected and the same process is performed, and the above process is executed for all pixels (step 5P40). After the processing for all pixels is completed, an average is taken for several diameters whose center coordinates fall within a certain range (step 5P41), and this is sent to the management computer 40 as the determined diameter (step 5P42).

FIG. 11(d) shows a specific example of the above tracking operation, and PO-PI3 in the figure is image data of "F" and "F". First, the pixel of interest PO is selected.

Next, the pixel PI corresponding to pixel A in the first map (a) is selected, and since the vector ff1Vx=I and Vy=I, the vector sum TVx-1 and TVy=I. Also, since the ratio of these is 1, Fig. 11(c)
Accordingly, the next analysis pixels A, B, and D are selected.

In this case, the pixel A and B at the pixel PI are "00", so the pixel P2 at the pixel position is "00".
is selected. The vector amount of this pixel is Vx=I
, Vy=0, so the vector sum 'I''Vx=1+
1=2, TVy=1+0=1.

Also, since these ratios TVy/'rVX= 1/2, from FIG. 11(c), the next analysis pixels are A, D, r
It becomes 3.

Therefore, pixel P4 located at position A with respect to pixel P2 is selected. At this time, although the pixel P3 is "FF", the pixel P4 located at the position of the pixel A is preferentially selected. In the same way, tracking is continued at pixel P12 until no subsequent points are recognized, and then it returns to pixel PO once again.
When the analysis of adjacent pixels A to H is completed, the process moves to the next pixel of interest PO.

FIG. 12 is a diagram for explaining the diameter calculation method in step SP38.

In this figure, three points on the arc are Q+(x+, Y+).

Let Q 2 (X2.Y2) and Q 3 (X3.y3). Also, line segments Q and Q.

The midpoint is Q4 (X4.F4), and the midpoint of line segment Q2Qs is Q
−(x,.

ys). Furthermore, the perpendicular bisector of line segments Q and Q is a
, the perpendicular bisector of line segments Q and Q3 is b. Thereby, the center coordinates (xo, yo) and the diameter d are determined as follows.

First, the slopes Da and Db of the line a and the line are orthogonal to the line segments Q, Q, , Q2Q3, so Da=
(X2 XI)/ (J2 Y+)・・・・・・(
3) D b-Cx3X2)/ (F3)'t)...
-(4) Also, midpoint Q4. The coordinates of Qs are )[, = (x+ + xy)/2, Y4= (y+
+yt)/2 ・--C5)X5= (x, +XI
)/2, ys = (Y2 + y3)/2−-
It is given by (6). Therefore, if we find the equations of lines a and b whose slopes are given by Da and Db and which pass through the midpoints Q, Q5, and set x=O, we can calculate the intercepts Ca of lines a and b using the following formula.
, Cb are determined.

Ca=-x4XDa+y+°°°°°°(7)Cb=X
s X D b+ Vs (8) Also, the center coordinate Q given as the intersection of two lines a and b. (xo
, yg) is xo=(Ca Cb)/(Da Db)...-(9
)yo: -DaX XOX C mini・・・・・・・・・
(IO). Furthermore, the diameter d is given as the distance between the point Q and the center coordinates.

Next, each process of color tone measurement and color tone determination will be explained with reference to FIGS. 13 and 14.

When the object to be inspected passes through the COD camera 4 and reaches the color difference meter 5, a proximity sensor (not shown) detects this, and the input contact group 36
, A/D conversion CP via the line control CPU 35a.
A color tone measurement command is given to O32a (step 5P5
1). When the start signal is output from the color difference meter 5 (the fourth
．． 5), the CPU 32a moves from step 5P52 to step 5P53, and performs the first data acquisition.

That is, the filters 54 B, 54 R, 54 in FIG.
Each color signal output from Y is taken in sequentially and A/D converted. And all the filters 54 B, 54 R, 5
When data collection from 4Y is completed (Step 5P54)
), 2nd time, 3rd time, etc. in the same way as above.
- Collect each color signal for the Nth time. When the Nth collection is completed (step 5P57), each filter 54B,
The average of N color signals of 54R and 54Y@ is taken (step 5P58), and the result (color tone data) is sent to the management computer 40. This transmission is performed via the communication CPU 33a. In addition, in step 5P54, if the start signals are not obtained in the correct order, steps 5P55 to 5P5 are performed.
Proceed to step 6 and send the error code to the management computer 40.

On the other hand, the management computer 40 that has received the color tone data or error code performs step SP6 in FIG.
At step 1, reception is detected, and at step 5P62, the receiver checks whether the received data is color tone data or an error code. In the case of color tone data, the process proceeds to step 5P63 to determine the color from the color tone data, and in the case of an error code, error processing is performed in step knob 5P64.

The above color tone determination is obtained by calculating the component ratio of each color. That is, the red, yellow, and blue color components are R and Y.

B, r=R/(R+Y+B) y= Y/(R+ Y + B) ・・・・・・(
I 1) Determine the proportions of each component r, y, b using the formula b=B/(R+Y+B), and calculate (if
Rate the magnitude of the value for r on a scale of 10 to 10.

In this case, the larger the value b/r, that is, the closer to black, the higher the score.

Next, measurement of the exposed proportion of the stem will be explained with reference to FIGS. 15 and 16. In this process, as shown in FIG.
, bi are sequentially read out and subjected to the processing shown in FIG. 16, which is executed by the image processing device 20.

Here, the extraction of the stem (green component) is performed under the following conditions.

(1) When the three data ri, gi, and bi are all larger than the set threshold, that is, when the luminance of the pixels corresponding to these data exceeds the predetermined threshold, they are removed as a blank page (background) as shown in FIG.

(2) When the values of the three data ri, gi, and bi are almost equal, it is also removed as a blank page.

(3) When green data gi is larger than either data ri or bi, that is, when the following conditions are met, it is determined that it is a stem.

(gi>ri) ・(gi>bi)---(12)(
4) Anything other than the above (1) to (3) is determined to be a bunch of grapes.

Then, the ratio P (%) of the stem to the bunch of grapes is determined by the following formula.

P・100XGcNT/(GCNT+BCNT)...
(13) Here, GCNT is the number of pixels determined to be a stem. BCNT is the number of pixels determined to be a tuft. FIG. 16 is a flowchart showing the flow of this stem exposure ratio measurement.

-J+'' ---Jhanri1-fu^-, hjikumt++t
After carrying out llll/shi, in steps 5P72 to 74, the above-mentioned determinations (1) to (3) are executed, and the number of green pixels G C
NT and the number of blue pixels BCNT (Step 5P7)
5.76). Then, data ri corresponding to the next pixel,
gi and bi are sequentially selected (step 5P78) and the same processing as above is performed, and when the processing for all pixels is completed (step 5P77), the exposure ratio of the stem is calculated using equation (13) (step 5P79). The results thus obtained are sent to the management computer 40 (step 5).
P80).

The management computer 40 stores the diameter data and stem exposure ratio data obtained by the image processing device 20, the color difference meter 5, and the CP.
The quality of the grapes is determined based on the color tone data obtained in U32a. In other words, each IO
Scores are given as a perfect score, and these are multiplied by weighting coefficients and totaled to give a total score of 5: excellent 1, excellent 2, excellent, average, and outside.
Evaluate in stages.

This evaluation result is sent to the line control CPU 35a via the communication CPU 33a. Then, when the cardboard box arrives at the sorting mechanism 7, the output contact group 37 corresponding to the above evaluation is activated, and the grapes are sorted.

[Effects of the Invention] As explained above, the present invention extracts the contours of an object to be inspected that overlap each other like the fruits of a grape, examines the shape, examines the color tone of the object, and synthesizes these. This method determines the grade of the inspected object and automatically selects the inspected object based on this, making it possible to always perform stable and objective evaluations without the intervention of the inspector's subjectivity as in the past. At the same time, it is possible to improve work efficiency and reduce the labor of the worker. Further, if the selection is performed by determining the mixing ratio of the object to be inspected and other objects, it is possible to perform the selection with even higher accuracy.

[Brief explanation of drawings]

FIG. 1 is a plan view showing the overall configuration of a sorting device according to an embodiment of the present invention, FIG. 2 is a block diagram showing the electrical configuration of the same embodiment, and FIG. 3 is a memory map of the storage section of the image processing device 20. , FIG. 4 is a block diagram showing the configuration of the color difference meter 5, FIG. 5 is a waveform diagram showing the output of the same color difference meter, FIG. 6 is a flow chart showing the overall operation of the same embodiment, and FIG. 7 is the image processing device. 8 is a flowchart showing the operation of preprocessing, FIG. 9 is a conceptual diagram showing the movement of data in preprocessing, FIG. 10 is a flowchart showing the operation of calculating the direct weight of grape fruits, Fig. 11 is a conceptual diagram for explaining the method of tracking continuous components of image data, Fig. 12 is a conceptual diagram showing a specific method for calculating the diameter, Fig. 13 is a flowchart showing the flow of color tone measurement, and Fig. 14 15 is a conceptual diagram for explaining a method for calculating the exposure ratio of a stem. FIG. 16 is a flow chart showing the flow of the calculation method. 4...CCD camera (imaging means), 5...
...Color difference meter, 23... Image processing CPU (extraction means, shape judgment means, object judgment means), 32a...
...CPU 140 for color tone determination...Management computer.

Claims (2)

[Claims] (1) An imaging means for taking images of a plurality of objects to be inspected that are located one on top of the other like the fruits of a grape; an extraction means for extracting the outline of each of the objects to be inspected from the output of the imaging means; a shape determining means for determining the size and variation thereof of the object to be inspected; a color difference meter that irradiates light onto the object to be inspected and obtains a signal corresponding to the color tone of the object from the reflected light; and an output of the color difference meter. A sorting device comprising a color tone determining means for determining the color tone of the object to be inspected, and sorting the object to be inspected based on outputs from the shape determining means and the color tone determining means. (2) an object determining means for extracting an object mixed in the object to be inspected, such as a stem of a bunch of grapes, from the output of the imaging means and outputting a mixing ratio of the extracted object and the object to be inspected; 2. The sorting device according to claim 1, wherein the inspection object is sorted based on the outputs of the shape determining means, the color tone determining means, and the object determining means.