JPH0415033B2 - - Google Patents

Description

[Detailed description of the invention]

[Industrial Field of Application] The present invention relates to a sorting method for sorting long side dishes such as cucumbers and eggplants into classes and grades based on the results of imaging using an imaging means. [Conventional technology] Long vegetables such as cucumbers and long eggplants are natural products, unlike factory-produced products, so their shapes and appearances vary widely. When grading wild vegetables, such as Japanese cucumbers, the sorting standards shown in Table 1 are used for the grading criteria that are quantified, and the grading standards that are not quantified are used. Regarding the standards (visual judgment standards established by each agricultural cooperative), the ones used are ones that illustrate objects with poor appearance as shown in (d) to (g) in Figure 1 using photographs, etc., and add explanations using written expressions. It is being Then, for the yellow cucumbers that have been automatically graded based on the sorting standards quantified in Table 1, those with poor appearance are lowered in grade by visual judgment by the sorter based on the grading criteria expressed in text. Currently, the company is trying to improve its product value. In other words, the elements of the criteria for grading the yellow lily that have not been quantified are (d) "white belly", (e) "bent neck" where the end is locally short, and (f) the end. Items such as "hang neck, narrow end" where the part is locally thin, "thick neck, thick end" where the end is locally thick, etc., and the selection standards are common throughout the country. "Thickness and curve"
Even if it is within the category of "excellent products", the above-mentioned Figure 1 (d),
If the appearance falls under any of the categories (E), (F), and (G), the quality of the Kiyuuri will be downgraded to "excellent, not excellent," and the quality of the Kiyuuri to be shipped will be reduced. The current situation is that we are striving to improve the value of our lineup of products. On the other hand, conventionally, fruits and vegetables are photographed by TV cameras.
Many technical documents have been published on methods and devices for measuring and sorting the outer diameter, length, and bending over the entire length using imaging devices such as CCD cameras and MOS type one-dimensional image sensor cameras. Applications of these techniques are known in Japanese Unexamined Patent Application Publication No. 54-145164. [Problems to be Solved by the Invention] However, the publicly known Japanese Patent Application Publication No. 54-145164 does not specifically meet the standards for sorting and packing fruits and vegetables (horticultural products) determined by each prefectural agricultural cooperative federation. This relates to a method of measuring and selecting the thickness and curvature over the entire length, which are numerically indicated grade determining factors, and the length, which is a class determining factor, and the above-mentioned non-numeric grade determining factors. has not been adopted, and the question of how to incorporate it remains unresolved, and facilities using this known method have been dissatisfied with the screening results despite the high facility costs. In other words, in this known product, if the thickness and bending are within the standard for excellent (for example, the curve is within 15 mm for Hidden M), a curved item that is curved over the entire length may be determined to be excellent. However, as shown in (e) in Figure 1, which should be downgraded from excellent to excellent, there is a drawback that even locally curved edges are judged as excellent. This shortcoming was due to the fact that the feature representing the shape of the end of the class was not incorporated as a factor in determining the grade (excellent, excellent, good, average). Furthermore, in the visual judgment performed by sorters regarding such grade judgment factors, it is undeniable that the judgment criteria differ slightly depending on the person, and there is a drawback that the visual judgment results vary due to individual differences. Furthermore, since the above-mentioned known method observes the entire surface of the lily and performs image analysis, it takes a long time to process, and the processing capacity cannot be improved. SUMMARY OF THE INVENTION An object of the present invention is to provide a method for sorting long side dishes, which eliminates these drawbacks and allows grade determination to be performed at high speed and with high accuracy without variation. [Means for Solving the Problems] In order to achieve the above object, the present invention images long side dishes as objects to be sorted conveyed on a sorting conveyor by an imaging means, and analyzes the images to determine the objects. In a sorting method that measures grade and class determination factors based on the shape of the sorted objects, determines the grade and class of the objects to be sorted based on the measurement results, and sorts the objects to be sorted by grade and class based on the determination results, The class of the object to be sorted is determined by measuring the length as the class determining element, and the class determining element is at least the width on the center line of each of the two ends, among the feature values representing the thickness, bending, and shape of both ends. This is a method for sorting long side dishes in which the grade of the object to be sorted is determined by measuring the bending angle of a straight line connecting three predetermined points sequentially from the center side. Even if the grade is determined only by measuring the thickness and the bend over the entire length, it is within the "Excellent" range, but at least the bend angle at both ends is measured among the feature quantities representing the shape of both ends. In this way, in the case of poor appearance at the edges, the grade is downgraded from "excellent" to "excellent", thereby making it possible to equalize the quality and increase the product value. [Embodiment] A preferred embodiment of the present invention will be described below with reference to the drawings. Figure 2 shows the main parts of a sorting device for sorting long side dishes, such as cucumbers.To give an overview, 1 is the bucket of the sorting conveyor, 2 is the material to be sorted, For example, Kiyuuri. Bucket 1 on which to place the cucumbers as the object to be sorted 2
The mounting surface is a black light-absorbing surface that hardly reflects light, and even if the surface is somewhat dirty, the dirt will not be imaged. 3 is an imaging device, and a television camera using an image pickup tube such as a vidicon, or a known camera using a CCD or MOS type solid-state imaging device is used. Reference numeral 4 denotes an illumination device, which illuminates the object 2 to be sorted using a strobe flash when the image pickup device 3 is an image pickup tube camera, and a lamp when the image pickup device 3 is a solid-state image pickup device camera. Reference numeral 5 denotes a switch for the bucket clock signal of the sorting conveyor, which outputs a clock signal that gives the timing to start imaging each time each bucket 1 reaches the imaging position. Reference numeral 6 is a rotary encoder which outputs the amount of travel of the sorting conveyor as a coded signal. In other words, it emits 1000 pulses for every 100 mm of movement of the sorting conveyor.
1 per 0.1 mm movement of the sorting conveyor
Outputs a signal at the rate of pulses. Reference numeral 7 denotes an image processing arithmetic unit, which includes a video signal slicer, a memory, an image analysis section, a standard value setting section, a calculation section, a sorting and sorting output section, a power supply section, and the like. In addition to configuring the sorting device as described above, the sorting standard of the material 2, that is, the class,
Set the grade determination element items as follows. (a) Length (class), (b) Thickness (class), (c) Curvature (class), (d) Belly white (downgrade), (e) Neck bend [bent angle at end] (downgrade) , (f) Hanging neck, thin end [thickness of the end]
(Downgrade), (G) Bottom thickness, neck thickness [end thickness] (Downgrade) The seven items (A) to (G) above are explained below with reference to the selection standard table in Table 2. Measure as follows. First of all, regarding the classification according to the length of the item (a) above, as shown in FIG.
is measured as length. Points a and b at both ends are, for example, when a and b are the width center points of the part where the thickness of both ends reaches a constant x dimension, and when a is the width center point of a constant y dimension part from both ends. , b, but the latter is preferred. By comparing this measured length dimension l with the class division value preset by a digital switch in the standard value setting section of the image processing arithmetic unit 7, the object to be sorted 2 is classified into LL, L,
It will be judged as one of the following classes: M, S, SS. Next, regarding the grading according to the thickness mentioned in item (b) above, as shown in Fig. 4, the middle part l is obtained by removing the part obtained by adding the dimension c from the points a and b at both ends of the object image 2'. ₁ , the thicknesses D ₁ , D ₂ , D ₃ , and _Do of the portions are measured at a plurality of locations, for example, n locations, for each small dimension l ₂ . These D ₁ , D ₂ , D ₃ , _and Do are measured in a direction perpendicular to the width center line (hereinafter referred to as the center line) s, and D ₁ ~
Detect the minimum and maximum dimensions from the D _o dimensions,
This value is compared with the grade division values preset in the standard value setting section, and the object 2 to be sorted is graded into the lower grade corresponding to either the size or the small value. That is, those of medium thickness are ranked as the best, Shu-grade, those that are slightly thicker or thinner are ranked as Superior, and those that are even thicker or thinner are ranked as Good. Experiments show that the distance between measurement points, i.e., the small dimension of l ₂
We were able to grade the thickness based on a good thickness standard of about 10 mm to 20 mm. Next, regarding the grading according to the curvature in the above item (c), each of the thicknesses measured with respect to the straight line w connecting the points a and b at both ends of the object image 2' as shown in FIG. Among the perpendicular lines v ₁ , v ₂ , v ₃ , and v _o drawn from the measurement center points s ₁ , s ₂ , s ₃ , and s _o , the maximum dimension v _nax is the bending dimension, and this value and the standard value setting section are The object 2 to be sorted is graded into the corresponding grade by comparing it with the preset grade division values for each length class. If the measurement results of the above two items (b) and (c) are rated in different grades (for example, thickness is excellent, bending is excellent or good, or the reverse combination of these), The grade belonging to the lowest among them is classified as the grade of the object to be sorted 2. Next, regarding the grade downgrading due to whiteness in the above (d), as shown in Fig. 6, the output level of the image signal of the object to be sorted is expressed as the outer shape value hν and the whiteness value Hν, and the area of the edge image is E and belly white area F are each measured, the ratio of F to E is calculated as G%, and this value is compared with the belly white downgrading category value preset in the standard value setting section. Sorted material 2 is determined to be a grade that is one grade lower than the grade that was graded according to items (b) and (c) above. Note that this downgrading based on blanks does not need to be used in the present invention. Next, regarding the grade downgrading due to the bent neck [bent angle at the end] in item (e) above, as shown in Figure 7, point a and
The angle at which the straight line j ₀ passing through the s ₀ point and the straight line k ₀ passing through the s ₁ and s ₀ points intersect, that is, the predetermined three points a, s ₀ , s ₁ on the center line s at the end are on the end side The bending angle θ _a of the straight line sequentially connected from the central part side on the opposite side (3 points are a, s ₀ ,
The obtuse angle formed by connecting s ₁ < supplementary angle of as ₀ s ₁ ) is measured as the neck bend. Also, on the symmetrical side, there are three predetermined points b on the center line s at the angle where the straight line j _n passing through points b and s _n and the straight line k _n passing through points s _o and s _n intersect, that is, at the end.
Bending angle θ _b of a straight line connecting s _n and s _o sequentially from the center side opposite to the end side (obtuse angle formed by connecting three points b, s _n , and s _o < supplementary angle of bs _n and s _o ) is measured as the neck bend. Then, the larger of the measured bending angles θ _a and θ _b at both ends is calculated as the neck bend angle θ,
This value is compared with the neck bend downgrade classification value preset in the standard value setting section, and if the value is found, the object to be sorted 2 is downgraded by one grade. That is, according to this grade downgrading method, if it falls under the downgrading classification value, the above (b) thickness, (c)
The grade that has been graded based on the two-item grade rating of bending is further downgraded to one lower grade, and then the grade is determined. For example, the item to be sorted 2, which is an "excellent" product that was rated as excellent in all of the grading items (b) and (c) above, has a neck bending angle θ in the item (e) that is lower than the standard value. If there is a large bend, the object 2 to be sorted is judged to have a poor appearance and is downgraded to "excellent" for grade determination. In addition, the material to be sorted 2, which is an "excellent" product that was rated as excellent in the grading items of (b) and (c) above,
If the neck bending angle θ in item (E) above is larger than the standard value, the object 2 to be sorted is judged to have a poor appearance and is downgraded to a "good product" and graded. Also, regarding the method of downgrading due to neck bending in item (e) above, according to experiments, i and i in Figure 7 are 20
By setting the neck angle θ to ~25 mm and approximately 20 degrees, we were able to obtain good appearance judgment results. Note that the set values of symbols i and θ in the figure are variable, so it is preferable to change the standard values as appropriate in relation to the market price depending on the product type and time. Next, regarding the grade downgrading due to the hanging neck and narrow end [thickness of the ends] in item (F) above, as shown in Figure 8, the thickness d ₀ at the predetermined points s ₀ and s _n at both ends. , d Measure _{the n} dimension perpendicular to the center line s and find the dimension
Let the thinner of d ₀ and d _n be the hanging neck and the thinner end d,
This value is compared with the downgrading classification value of the vine neck end fineness preset in the standard value setting section, and if it is applicable (if it is thinner than the standard value), the material to be sorted 2 is
The grade that has been graded under (b) and (c) will be determined to be a grade that has been downgraded by one grade. According to experiments, it was possible to obtain good appearance judgment results by setting the classification value to about 2/3 of the thickness D. Next, regarding the bottom thickness and neck thickness [thickness at the ends] in item (g) above, as shown in Figure 9, the thicknesses at point s ₀ and point s _n are D ₀ and D as in item (f) above. _n in a direction perpendicular to the center line s, and the larger of the dimensions D ₀ and D _n is defined as the bottom thickness and neck thickness D, and this value and the bottom thickness preset in the standard value setting section, When compared with the downgrading classification value of the neck thickness, if applicable (if it is thicker than the standard value), the material to be sorted 2 will be downgraded by one grade with respect to the grade rated according to the above (b) and (c). It will be judged according to the grade. In addition, the downgrading of the grade is based on the judgment factor items (E), (F), and (G).
If one of the items corresponds to each of the individual downgrading category values preset in the standard value setting section, or if multiple items apply (for example, two items of "bent neck" and "thick butt"), the downgrading will be by one grade only. and the grade is determined. Furthermore, the measurement in this description refers to measurement by imaging analysis. The sorting device used in the present invention can be a device combining 1 to 6 in FIG. , it is preferable to arbitrarily design and use it according to the size. [Effects of the Invention] As described above, the present invention sets the measurement points at a, s ₀ , s ₁ , s 2 , along the width center line of the imaged object to be measured 2 _' .
Since the measurement is performed by determining s ₃ ...s _o , s _n , b, etc., the image analysis measurement speed is fast, and the processing capacity can be further improved compared to the conventional method. Further, the present invention determines the class by measuring the length when sorting long side dishes as objects to be sorted, and determines the class by measuring the length, and at least both ends of the feature values representing the thickness, bending, and shape of both ends. Since the grade of the material to be sorted is determined by measuring the bending angle of a straight line connecting three predetermined points on the width center line of each section sequentially from the center side, the thickness and overall length are used as grade determination factors. Conventional methods that measure the bending of the arm may not be able to determine it, or the appearance of both ends may be poor.
Since the bending angle as the bending of the neck is automatically taken into account when determining the grade, there is no dispersion in the determination results as would be the case with visual determination, and the grade can be determined with high accuracy. The quality of the sorted items can be made uniform and the product value can be increased.

【table】

【table】 [Brief explanation of drawings]

The drawings all show embodiments of the present invention, and FIGS. 1(a) to (g) are explanatory diagrams showing various shapes of objects to be sorted, FIG. 2 is an explanatory diagram of the main parts of the sorting device, and FIG. 3,
Each figure in figures 4, 5, 6, 7, 8, and 9 shows the length, thickness,
An explanatory diagram of each measurement procedure for bending, belly white, neck bend, hanging neck and thin butt, thick neck and thick butt is shown. DESCRIPTION OF SYMBOLS 1...Bucket, 2...Object to be sorted, 3...Imaging device, 4...Lighting device, 5...Bucket clock signal switch, 6...Rotary encoder, 7...Image processing calculation device, l... ...length, a,
b...Measurement point, s...Width center line, D, d...Thickness, v...Bending dimension, G...White belly ratio, E, F...
...Image area, θ, θ _a , θ _b ...Bending angle, s ₀ , s ₁ ~ s _o ,
s _n ...Each measurement point.

Claims (1)

[Scope of Claims] 1. An imaging means captures an image of long side dishes as objects to be sorted conveyed on a sorting conveyor, and measures grade and class determination factors based on the shape of objects to be sorted by analyzing the images. A sorting method in which the grade and class of the objects to be sorted are determined based on the measurement results, and the objects to be sorted are sorted by grade and class based on the determination results, wherein the objects to be sorted are determined by measuring the length as the class determination element. The class of the object is determined, and among the feature values representing the thickness, curvature, and shape of both ends as the grade determination factors, at least a straight line connecting three predetermined points on the width center line at each end sequentially from the center side is used. A method for sorting long vegetables, characterized by determining the grade of the object to be sorted by measuring the bending angle. 2. The long side dish according to claim 1, wherein the feature amount representing the shape of the both ends is the bending angle and the thickness at predetermined points S ₀ and S _n of both ends. selection method. 3 Thickness and bending as the grade determination factors are as follows:
2. The method for sorting long side dishes according to claim 1, wherein the measurements are performed at a plurality of predetermined locations in the middle portion obtained by removing a predetermined dimension (y+c) from each end. 4. The length of the long side dish as set forth in claim 1, wherein the length as the class determination element is a straight line distance between the width center points a and b within a predetermined distance from both ends, respectively. Sorting method.