JPH0360774A - Color detection type device for discriminating kind of leaf tobacco - Google Patents

Abstract

PURPOSE:To automatically discriminate the kinds of leaf tobacco by providing a color measuring means, an input means for inputting the information to indicate the measuring section and a memory means. CONSTITUTION:A color sensor 1 cast illumination light to a material to be measured and outputs the measured value based on the reflected light as digital data. The color space value from the color sensor 1 is taken in via an interface 3 by a decision control means 2 which makes various kinds of calculation and displays the result of the decision to a display means 4. The data indicating the leaf taking position of the leaf tobacco, the data indicating the measuring section and the data indicating criteria ar inputted from an input means 5. An output means 28 and a control means 29 consist of the functions of the control program previously stored in the memory. The ranking of the leaf tobacco is easily executed in this way.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention provides a leaf tobacco sorting device suitable for use in grading the quality of harvested and dried leaf tobacco.

[Conventional technology]

Traditionally, the quality of leaf tobacco has been determined by the position of young leaves, ripeness,
They are evaluated based on the quality factors of color, texture, and mesophyll, and are classified by the naked eye as shown in Table 1 below.

Prices are then set corresponding to each purchasing category (categories marked with "○").

Leaf Tobacco Purchasing Classification Table Table 1 Among the purchasing classifications in the table above, P-type leaf tobacco, which has discolored leaves, is a type of leaf tobacco that is known as a rapidly drying leaf, and is rapidly dehydrated after yellowing during the post-harvest drying process. The natural browning of leaf tobacco has not been sufficient and a yellowish tinge remains. In addition, among the poorly ripened middle leaves (upper leaves, middle leaves, and lower leaves), blurred leaves with a dull whitish hue and no luster in appearance occur together with rapidly drying leaves, so blurred leaves are also P type. It is rated as such. These P-type leaf tobaccos have problems such as a bad tobacco taste and the fact that the flavoring effect is not fully exerted during the tobacco manufacturing process, so they are different from regular types in terms of price. For this reason, special attention is paid to the P type rating.

However, since the P-type leaf tobacco described above is similar in hue to the ordinary A-type leaf tobacco, there is a problem in that it is difficult to grade even an expert using the conventional visual grading method.

An object of the present invention is to enable automatic discrimination of the type of leaf tobacco based on the color of the specimen.

[Means to solve the problem]

The color detection type leaf tobacco sorting device of the present invention, which was made to solve the above problems, detects the color of a sample and converts it into a color space of a color system expressed by a brightness index of 1 and a perceptual color index of 2. a color measuring means for outputting coordinate information; an input means for inputting information indicating the measurement site of the leaf tobacco sample; a storage means for storing information indicative of the characteristic boundaries of in association with the type of leaf tobacco, information indicative of the characteristic boundaries stored in the storage means and coordinate information from the color measurement means; From a calculation means for calculating the distance between the measurement point of the leaf tobacco and the characteristic boundary in the color space, and a discriminant function that is set in advance by associating the attribute probability for identifying the type of leaf tobacco with the distance from the characteristic boundary, an attribute determining means for outputting an attribute probability corresponding to the distance obtained by the arithmetic means; A determination means for determining the type of leaf tobacco to be measured from the stored attribute probabilities of a plurality of parts based on a preset determination criterion by associating the combination of attribute probabilities of the measurement parts with the type of leaf tobacco. Features.

[For production]

In the color detection type leaf tobacco sorting device of the present invention, information indicating characteristic boundaries of the leaf tobacco distribution area is stored in the storage means, and the calculation means combines the information indicating the characteristic boundaries and the coordinate information from the color measurement means. Based on this, the distance between the measurement point of the leaf tobacco and the feature boundary in the color space is determined. The attribute determination means outputs an attribute probability corresponding to the distance obtained by the calculation means from a discriminant function set in advance, and the determination means outputs the attribute probability obtained by the attribute determination means specified by the input means. The type of leaf tobacco to be measured is determined from the attribute probabilities of a plurality of parts stored in association with the measurement parts and based on preset determination criteria.

The color of leaf tobacco is distributed over a certain width on a curve in a color space such as the Lsa*b* color system, and P type leaf tobacco is mainly distributed on the outer peripheral side of this distribution area.

Therefore, for example, if the boundary between the distribution areas of regular leaf tobacco and 9541 leaf tobacco is defined as a characteristic boundary, based on the distance between this characteristic boundary and the measurement point, 254
It is possible to obtain attribute probabilities for identifying the type of leaf tobacco, and furthermore, by determining the type by combining the attribute probabilities of multiple measurement parts of leaf tobacco, the entire leaf tobacco can be considered even if the type differs depending on the part of the leaf tobacco. 2541 leaf tobacco, etc. can be graded.

[Example] Hereinafter, the present invention will be described in detail using a type of leaf tobacco as an example.

Figure 10 is a diagram showing the color distribution of leaf tobacco in the color space of the Lsa* b* color system, and Figure 11 is a diagram showing A-A in Figure 10.
2 is a diagram showing a cross section, and the distribution of leaf tobacco shows a constant color change path within the color space corresponding to the young leaf position shown in Table 1 above. Ordinary tea leaf tobacco is distributed along a curved transition path in the order of lower leaf, middle leaf, golden leaf, wood leaf, upper leaf, and upper leaf, and 2541 leaf tobacco is concentrated in the outer region of ordinary tea leaf tobacco. It is especially widely distributed between Kinshu and Konoha.

The present invention provides L7 color for multiple parts of leaf tobacco.
Measure the data of the lightness index L9 and the perceived chromaticity index a''', bl (hereinafter referred to as color space value) in the color space of the a*b'' color system, and calculate the data in the color space corresponding to this color space value. 2541 based on the distance (color difference) between the measurement point and the distribution area
Leaf tobacco can now be identified. In addition, as shown in Figure 11, there are 254 tree leaves (tree leaves, upper leaves, and heavenly leaves).
The regions of single-leaf tobacco are mesophyll (lower leaf, middle leaf, and golden leaf).
In the following examples, the 2541 leaf tobacco is distinguished by distinguishing the young leaf position of the leaf tobacco, that is, the woody leaf type and the middle leaf type.

Next, the determination method in the embodiment will be explained.

(Discrimination method) First, for the color space of the l, *a*b0 color system explained with reference to FIG. Two types of three-dimensional discriminant spaces are set up. And the opening ceremony (
By the coordinate transformations in 1) and (2), the distribution areas of woody-leaf and medium-leaf tobacco in the above color space are respectively assumed in each discriminant space, and the distribution areas in each discriminant space are defined by the characteristics of the ellipse, which will be explained later. Preset by boundaries. Also,
As will be explained later, the color space values from the color sensor that detects the color of leaf tobacco are also coordinate-transformed using formula (1) or (2) depending on the young leaf position of the specimen (woody leaf type or middle leaf type), and converted to the young leaf position. Each of 2541 leaf tobaccos is discriminated within the corresponding discrimination space.

When the distribution of multi-leaf tobacco (woody-leaf type and medium-leaf type tobacco) in the color space is transformed into the two types of discriminant space as described above, the center line of the distribution area of regular tea-leaf tobacco becomes as shown in Figure 6 in the case of tree-leaf type tobacco. As shown, it can be approximated by an arc and a straight line, and
In the case of mesophyllosis, it can be approximated by an arc in the discriminant space as shown in FIG. Then, the center line of the distribution area of the tree-leaf type becomes a circular arc shown by the following equation (3) and a straight line, and the center line of the distribution area of the meso-leaf type becomes a circular arc shown by the following equation (4). Note that in FIGS. 6 and 7, the coordinate axis z3 is perpendicular to the coordinate axis ZlyZ2.

(Kinoha type) (Middle leaf type) z,! ＋zt! = 10” 9 Z3=0
On the other hand, the distribution area of leaf tobacco is around each center line as shown in Figure 1O above, but its width is not constant, and 2541 leaf tobacco exists in all areas around each center line. Therefore, in the case of a mesophylloid, as shown in Figure 7, the center line is z, z, and four divisions (■ ~
(2)) and set criteria for judgment for each of the divided categories.

In the case of mesophyllosis, 2. in Figure 7. Angle counterclockwise around the origin with the positive direction of the axis as the reference (
4) When classified according to the following, the distribution of leaf tobacco is as shown in skin surface 2.

Table 2 As shown in Table 2 above, 2541 leaf tobacco is categorized as ■～■
Therefore, in the determination process, first, it is determined to which category the measurement point in the discriminant space corresponding to the measured color space value belongs in terms of Z, Z, and plane, and if it belongs to the categories ■ to ■, 2541 Identify leaf tobacco. Note that the coordinates (Z+?22 *Z3) of the measurement point in the discriminant space are obtained by the coordinate transformation described above, and the angle α indicating the division in Z, Z, and the plane is α=jan-'('')... ...
・(5) Obtained by l. Then, the classification is determined by comparing this angle α with the angle ranges corresponding to the classifications (■) to (■), respectively.

As a criterion for each classification of 2541 leaf tobacco in categories ■ to ■, a ZaZ* coordinate system perpendicular to this center line is set along the center line of categories ■ to ■, and the boundary of the distribution area of regular tea leaf tobacco is The equation of the ellipse representing the feature is set in the z, Z= coordinate system. The origin of this Z23 coordinate system is on the center line, the Z4 coordinate axis is within the Z plane of the discriminant space, and the z3 coordinate is the same as the Z coordinate of the discriminant space.

The following equation shows the equation of the ellipse in sections 1 to 2, and its graphs are shown in FIGS. 9(a) to 9(C). In addition, θ in the following formula is the slope of the long sleeve of the ellipse, and 2. The angle is rotated counterclockwise around the origin with the positive direction of the axis as the reference.

Discrimination between regular tea leaf tobacco and 2541 leaf tobacco is performed based on the distance (color difference) from this characteristic boundary to the measurement point, using the ellipse for each category set as described above as a characteristic boundary in the discrimination space.

Note that the above distance L'dX is determined as follows.

First, the coordinates of the measurement point (zl, z2Z3) in the discriminant space are transformed using the following equation (6), and the measurement point Za Z: 1
Find the coordinates (Z4+Zl) on the surface.

23 = Zs t Zs = 1 piece; "'R""
(6) However, R is the radius of the center line (R=10), and the distance X as shown in the following equation (7) is determined by the length of the perpendicular line hanging from this point (Z49 Za) to the feature boundary.

Note that the value of distance X is negative inside the feature boundary and positive outside the feature boundary.

Now, the equation of the ellipse of the feature boundary is as follows: The measurement point is ('1an-Z+o), and the intersection of the ellipse and the perpendicular line hanging from the measurement point to the ellipse is (Z4P -23F).
Then, the distance X can be calculated using the following formula.

x" Lp Ln') 10 (ZB Z31
1) """ (7) However, z411' =
Z4El cosθ−Z3I, SinθZzo' =
Zso sinθ+ Z:10 cosθ
Z4F is calculated from the following quartic equation using the well-known Ferrari formula, a・(24F) ' +b・(Lp) ''+c4Lp)
2+d-(Z4P) +e=0 Also, 23P is determined by the following formula.

Next, in the case of tree leaves, see 2. in Figure 6. 2541 leaf tobacco exists only in the positive direction of the axis.

Also, 2. in the discriminant space. Since the distribution area of leaf tobacco around the axis is small, the determination process for regular tea leaf tobacco and 2541 leaf tobacco is performed using an ellipse on the Z, zz plane of the discrimination space as a feature boundary, as shown in Figure 8 and the following equation. , based on the distance (color difference) from this feature boundary to the measurement point.

Tatami, 0≦2゜ Note that the above distance X is determined as follows.

That is, the distance X as shown in the following equation (8) is determined by the length of the perpendicular line hanging from the measurement point (Z+yZz) in the discrimination section to the feature boundary.

Now, the equation of the ellipse of the feature boundary is as follows: The measurement point is (ZID-ZZD), and the intersection of the ellipse and the perpendicular line hanging from the measurement point to the ellipse is (ZIP, Zzp).
Then, the distance X can be calculated using the following formula.

X= Zip ZID + Zzp
Zto·=−(8) Note that ZIF is determined from the following quartic equation using the well-known Ferrari formula.

a・(ZIF) '+b・(Zip)'+c・(ZIF
) ”++L (ZIF) + e= 0 Also, zp is determined by the following formula.

As described above, the distance X to the characteristic boundary is determined for the medium-leaf type or the wood-leaf type, and P-quib leaf tobacco is determined based on the value of this distance X.

By the way, some leaf tobacco, even leaf tobacco,
There may be different types depending on the base, center, and tip of the leaf, and there are two types of leaf tobacco, one leaf tobacco or a bunch of leaf tobacco (a handful of leaf tobacco sampled for discrimination). There may be a mixture of P type parts. For this reason, when a bunch of leaf tobacco has uneven color, the current grading by graders is to grade each part of the leaf tobacco, as shown in skin table 3, and judge each part comprehensively. We are doing the grading.

Furthermore, as shown in Table 3, the degree of "Normal type-likeness" and "P-type-likeness" is evaluated for each part, and based on each evaluation result, the entire East It is rated as normal type or P type. In addition, in the skin surface and the following description, "N", "Light N", "Light P",
"Medium P" and "Heavy P", "N" is normal type, "light N"
is a normal type closer to P type, "light P" is a mild type of P,
"Medium P" represents a severe P type, and "severe P" represents a severe P type, each representing an attribute indicating the degree of "P type-likeness."

Table 4 For this reason, as shown in Figure 3, the distance from the characteristic boundary
A discriminant function indicating the degree of each attribute of N, light N, light P, medium P, 9 heavy P is set as a variable, and from this discriminant function, the probability that the measurement part has each attribute (hereinafter referred to as attribute probability) is set. )
I am trying to calculate.

less than, The discriminant functions of the tree-leaf type and meso-leaf type are shown, That graph, Figure 4 shows the leaf system. Figure 5 shows the mesophyletic system. (shown in al~(C).

(Discriminant function of tree-leaf type) (Discriminant function of meso-leaf type) Classification ■〉 k'zt(X) = PZI(X) -2,40≦X〈 0.72 Pz-(x) = -0,575・(× −1,02) −0,72≦χ〈 1.02 Category ■〉〈Category ■〉 PX3(X) 1.515・(X +0.54 ) −1,20≦X <−0,54 When the attribute probability is determined using the above discriminant function, the attribute of the measurement part (N) that has a large meat amount with the attribute probability is determined.

One of the five categories: light N, light P, medium P, and IP).

For example, as shown in Table 5 below, if the distance L'd x is in the distance interval CM, , M, It is an attribute of the part.

(Hereinafter, blank space) Table 5 Then, depending on the combination of attributes of each measurement site, the grading in Table 6 below is based on the rules, and the preset grading judgment values (P,
75, 50, 25. Find N).

(Hereafter, margin) Table 6 however, Judgment value of the above rating (hereinafter referred to as Rating judgment It's called value.

) indicates the percentage of grading tools that were evaluated as P type when a large number of graders graded a bunch of leaf tobacco for each combination of parts and attributes, and shows the following content.

P: A bundle of leaf tobacco that all the appraisal tools rate as P type 75: A bundle of leaf tobacco that 75 out of 100 people rate as P type 75: 75 out of 100 people rate Leaf tobacco in bundles of combinations that are rated as P type 507 Out of 100 people, 50 people rate leaf tobacco in bundles of combinations that are rated as P type Leaf tobacco N: A bundle of leaf tobacco that all the grading tools rate as normal. Grading judgment value (P, 75, 50.25.
Once one of N) is determined, the rating judgment value is compared with the judgment standard to determine the P type. In this embodiment, if the rating judgment value is P or 75, it is determined to be P type, and N
, 25 and 50, it is determined to be normal type.

Next, the configuration of the device of the embodiment will be explained.

(Configuration of Apparatus) FIG. 1 is a block diagram of an embodiment of the present invention, in which ordinary tea leaf tobacco and P type leaf tobacco are determined based on the above-described determination method.

In the figure, 1 is a color sensor that irradiates an object to be measured (leaf tobacco) with illumination light and outputs a colorimetric value as digital data based on the reflected light. The data (color space value) of the perceived chromaticity index a9, b" is output. As this color sensor, for example, "Color Difference Meter CR-, 110J manufactured by Dan Norta may be used.

Reference numeral 2 denotes a determination control means composed of a microcomputer or the like, and the functions obtained based on a control program stored in a memory (not shown) are shown in a functional block diagram.

The determination control means 2 takes in the color space values from the color sensor 1 via the interface 3, performs various calculations, and displays the determination results on the display means 4. In addition, data indicating the position of young leaves of the tobacco leaf to be measured (hereinafter referred to as young leaf position data), data indicating the measurement site (hereinafter referred to as measurement site data), and data indicating criteria are input from input means 5 such as a keyboard. be done. These data can be obtained, for example, by displaying the young leaf position and measurement site in a menu format on the display means 4, using a keyboard on the input means 5 to operate a predetermined key, and identifying the code of the operated key. It can be entered by

In the determination control means 2, the storage means 21 is configured in the memory of a microcomputer, and is based on the coordinate transformation equation (
1) Equation 9 (2), data on the angular range indicating each category ■ to
Equation (6) to obtain the coordinate value of Z, z = plane from z 9 z,, ), Equation (7) and Equation (8) to obtain the distance X from the measurement point to the feature boundary, the discriminant function to obtain the attribute probability, For the ratings shown in Table 6, data representing rules in a table format are programmed and stored. The storage means 21 also stores young leaf position data and measurement site data input from the input means 5.

The color space value reading means 22, the coordinate conversion means 23, the category discrimination means 24, the distance calculation means 25, the attribute judgment means 26, the rating means 27, the output means 2, and the control means 29 are stored in advance in the memory of a microcomputer or the like. The control means 29 associates the functions of each means based on a determination flow described later.

When the color space value reading means 22 detects that the measured value has been determined by the color sensor 1 based on the interrupt signal from the color sensor 1, the color space value reading means 22 reads L*, a output from the color sensor 1.
The m and b* signals are taken in through the interface 3, and the L's, a* and b* signals are output as numerical data.

When the young leaf position data is stored in the storage means 21, the coordinate conversion means 23 converts the data from the color space value reading means 22 based on the above-mentioned equation (1) or equation (2) corresponding to the young leaf position indicated by this data. Coordinate transformation is performed on the numerical data to generate coordinate data ZIt2+Zff in a tree-like or meso-leaf discriminant space.

The classification determining means 24 reads the young leaf position data in the storage means 21, and in the case of mesophyll, it determines z based on the above-mentioned formula (5).
, the angle in the Zz plane is determined, and the angle range of the mesophyletic division stored in the storage means 21 is compared with the angle determined by the calculation means 24 to determine the measurement point (ZI * Z
l, Z: l). Note that when the measurement point corresponds to a classification, the attribute probability of N is set to ``°1''.Also, when zl is negative in the case of tree-leaf type, the attribute probability of N is set to ``'l''.

The distance calculating means 25 calculates the above equation (6) in the case of mesophyllis.
Based on z, coordinate data 24 on the 23 plane.

z3 is determined, and when the classification determining means 25 determines that it is one of the classifications ■ to ■, the characteristic boundary of the determined classification is selected from the storage means 21 and the selected characteristic boundary and z4. Find the distance X from z. In addition, in the case of tree leaves, storage means 2
The distance X is calculated from the characteristic boundary of the true leaf system stored in 1 and Ziz2 obtained by the coordinate conversion means.

The attribute determination means 26 uses the distance X calculated by the distance calculation means 25.
The attribute probability is determined from the discriminant function stored in the storage means 21, and the attribute showing the maximum attribute probability for one measurement site is determined and temporarily stored in the storage means 21.

The grading means 27 reads the attributes of each part stored by the attribute discriminating means 26, and based on the combination of the attributes, the grading stored in the storage means 21 refers to the rule data and determines the grading judgment value. The determination criteria stored in the storage means 21 are compared with the rating determination value, and the P type determination result is output to the output means.

The output means 28 outputs display data indicating the determination result to the display means 4, and the determination result is displayed on the display means 4.

FIG. 2 shows that the determination control means 2
This is a flowchart showing the determination process performed in step S1), in which, when the control program is started, input of young leaf position data from the input means 5 is monitored (step S1), and when the young leaf position data is input, this input is The young leaf position data is stored in the storage output 21 (step S2).

When the color space value reading means 22 detects that the measured value of the color sensor l has been determined based on the interrupt signal from the color sensor 1, the color sensor l outputs L",
, bo signals are taken in through the interface 3, and the Lm, a*, b signals are temporarily stored as numerical data (Step S,), and the input of measurement site data from the input means 5 is monitored. (Step S4), when the measurement site data is input, the numerical data is associated with the measurement site data and stored in the storage means 21 (Step S).
.

Next, determine whether the tree is leafy or mesophyllous (step S).
, control is performed as follows depending on each case.

(In the case of tree-leaf system) The coordinate transformation means 23 transforms the numerical data stored in the storage means 21 according to equation (1) to obtain the coordinate data Zl, zz, z3 of the discriminant space (step 37),
Determine whether the ZIO value is negative (step S
,). Then, if Zl is negative, the attribute probability that the attribute of the part corresponding to this measurement point is N is set to "1" (Step 39), and if Zl is not negative, the distance calculation means 25 A distance X from the feature boundary is calculated (step 5I0).

(In the case of mesophyletic system) The coordinate transformation means 23 transforms the numerical data stored in the storage means 21 using equation (2) to obtain coordinate data Zly! of the discriminant space. , Z3 is generated (Sterap S,).

Next, the classification determining means 25 calculates equation (5) from zl, z.
) to find Z, Z, and the angle α in the plane, and further calculate coordinate data z4 . on the Z3z4 plane based on equation (6). z,
(Step S1□). Then, the angle range of the division stored in the storage means 21 and the obtained angle α are compared to determine the division of the point (z, , Zry Z3) (
Step SI:l).

As a result of classification, if the measurement point belongs to category ■ (65
<α<315), set the attribute probability that the attribute of the part corresponding to this measurement point is N to "1" (Step 5I4)
, when the measurement point belongs to any of the categories ■ to ■, the distance calculation means 25 calculates the distance X from the feature boundary corresponding to the category to which it belongs (steps Sls, S16?).
S+t).

In both cases of tree-leaf type and meso-leaf type, calculating the distance X (
Step S1゜t 5ISy S+bt 5at),
The attribute determination means 26 calculates the attribute probability of the measurement part from the predetermined discriminant function corresponding to each classification of woody type and mesophyll type stored in the storage means 21 (step S ls
).

Setting the attribute probability "1" of N (step S).

, S, ) or calculation of the above attribute probability (step S,
) is completed, the attribute indicating the maximum attribute probability is temporarily stored in the storage means 21 in association with the data indicating the measurement site (step 519).

Then, the above flow is repeated while determining whether the attributes of all parts (base, center, and tip of the leaf) have been determined (step S2゜), and when the attributes of each part are stored in the storage means 21, the grading The means 27 reads out the stored attributes of each part, and based on the combination of the attributes, the rating stored in the storage means 21 refers to the data of the rules to determine the rating judgment value (step Sz+) and stores it. Compare the judgment criteria stored in the means 21 with the rating judgment value and output the P type judgment result to the output means (step S
2□). Then, the determination result is displayed on the display means 4 by the output means 28.

In the above embodiment, the grading judgment value is determined according to the rules for grading according to the combination of attributes of each part, but the grading judgment value shown in Table 6 above is "N" → "25" → "
50” → “75” → rp.

may be expressed as continuous values.

For example, a determination expression as shown in the following equation (9) is set.

D=α, Σ(P,,) +β, Σ(Pz=)+7・Σ(
Pffi) + δ・Σ(P4i) +ε・Σ(Psi)・
...(9) Then, as in the above embodiment, N for each part.

Find the attribute probabilities of light N, light P, medium P and double P, and add the attribute probabilities of each part for each same attribute (each summation term of the discriminant),
A determination value (D) is calculated by assigning a weighting coefficient to each.

In addition, as a criterion for determining regular tea leaf tobacco and 2511 leaf tobacco, for example, a darkness value Ds such as the boundary between "75" and "N" in the above example is set in advance, and when DADs, it is determined as regular tea leaf tobacco. and 25 when Ds≦D
It is determined to be 11 leaf tobacco.

As described above, by measuring the color of leaf tobacco with a color sensor, it is possible to identify 2511 leaf tobacco.Furthermore, by measuring multiple parts of leaf tobacco, it is possible to identify leaf tobacco in which both ordinary type and P type are mixed. Leaf tobacco as a whole can be graded as P type.

Further, in this embodiment, by inputting data on the position of young leaves of tobacco leaves, it is also possible to discriminate 2541 leaf tobacco, which is a woody leaf type that has a similar hue to medium leaf type regular brown leaf tobacco.

〔Effect of the invention〕

As explained above, according to the color detection leaf tobacco type device of the present invention, information on feature boundaries indicating the distribution area of leaf tobacco in a color space such as the L′″a*b* color system is stored,
Based on the distance (color difference) between the measurement point in the color space of leaf tobacco measured by a color sensor, etc. and the memorized characteristic boundary, the attribute probability for identifying the type of leaf tobacco is calculated, and the attribute probability is determined by measuring at multiple parts of the leaf tobacco. Since the type of leaf tobacco is identified based on a combination of multiple attribute probabilities, it is not only possible to easily grade regular tea leaf tobacco and 2511 leaf tobacco, which have similar colors, but also a mixture of regular and P type leaf tobacco. It is also possible to grade P-type leaf tobacco as a whole.

[Brief explanation of drawings]

Fig. 1 is a block diagram of an embodiment of the present invention, Fig. 2 is a flowchart related to the embodiment, Fig. 3 is a diagram explaining a discriminant function in the embodiment, and Fig. 4 is a graph of a tree leaf discriminant function in the embodiment. Figure 5 is a diagram showing the graph of the discriminant function of the mesophyll system in the example, Figure 6 is a diagram showing the center line of the distribution area in the discriminant space of the tree leaf system in the example, and Figure 7 is the graph of the discriminant function of the mesophyll system in the example. A diagram showing the center line of the distribution area in the discriminant space of the mesophyll family in the example, Figure 8 is a diagram showing an example of an ellipse of the characteristic boundary of the tree leaf family in the example, and Figure 9 is a diagram showing the feature boundary of the mesophyll family in the example. FIG. 1O is a diagram showing the distribution state of leaf tobacco in the color space according to the present invention. FIG. 11 is a diagram showing a cross section of the distribution area of FIG. 1O. ■... Color sensor, 2... Judgment control means, 3... Input means, 4... Display means.

Claims (1)

[Scope of Claims] Color measuring means for detecting the color of a specimen and outputting coordinate information in a color space of a color system expressed by a brightness index of 1 and a perceptual color index of 2, and measurement of leaf tobacco as a specimen. an input means for inputting information indicating a part, and information indicating characteristic boundaries within a distribution area of the leaf tobacco in the color space for a plurality of leaf tobaccos classified in advance based on a plurality of types, in correspondence with the type of the leaf tobacco; and determining the distance between the measurement point of the leaf tobacco and the characteristic boundary in the color space based on the information indicating the characteristic boundary stored in the storage means and the coordinate information from the color measuring means. A calculation means outputs an attribute probability corresponding to the distance obtained by the calculation means from a discriminant function that is set in advance by associating the attribute probability for identifying the type of leaf tobacco with the distance from the feature boundary. an attribute determining means for storing the attribute probabilities obtained by the attribute determining means in correspondence with the measurement parts specified by the input means, and storing the attribute probabilities obtained by the attribute determination means in correspondence with the type of leaf tobacco and the combination of the attribute probabilities of the plurality of measurement parts. What is claimed is: 1. A color detection type leaf tobacco type device comprising: determination means for determining the type of leaf tobacco to be measured from stored attribute probabilities of a plurality of parts based on determination criteria set in advance.