JPH0370086A - Shape recognizing device - Google Patents

Abstract

PURPOSE:To surely extract the feature parts like a node, etc., by providing a light irradiating means which obtains an optical section shape of an object to be recognized and an irradiating position moving means which moves the light irradiating position in the extending direction of the object to be recognized at a prescribed interval. CONSTITUTION:An object 3 to be recognized is irradiated at a prescribed angle with the 1-dimensional light via a light irradiating menas 4 in the direction approximately orthogonal to the extending direction of the object 3. Thus an optical section form of the object is obtained and photographed by an image pickup device 6. Under such conditions, the object 3 and the means 4 are relatively moved by an irradiating position moving means 1 and the position of the optical section form of the object 3 is moved at a prescribed interval. At the same time, a width detection means 18 detects the edge position of the object 3 based on the optical section shape of the image data and obtains the width of the subject 3. The feature detection means 16, 19 and 20 detect the feature parts of the object 3 based on the width changes of the object 3. In such a constitution, the feature parts like the seedling nodes, etc., can be surely extracted despite the same color or the change of the stem diameter of a plant.

Description

[Detailed Description of the Invention] [Objective of the Invention] (Field of Industrial Application) The present invention is a shape recognition method for extracting locally swollen parts of a cylindrical body, such as the nodes of a stem of a seedling, such as a potato. Regarding equipment.

(Prior Art) In recent years, a method called tissue culture, in which seedlings of agricultural and horticultural crops are grown in large quantities in an artificially controlled sterile environment, has been studied. This involves creating Coulomb seedlings by cutting out the growing points, or nodes, in the plant that have the most proliferative activity and culturing them.

Therefore, when culturing Coulomb seedlings by such a method, it is required to automatically extract the nodes from the Coulomb seedlings and cut out the node portions. By the way, there are, for example, the following techniques for extracting a specific part of a plant from image data obtained by imaging the plant. That is, when the object to be extracted is, for example, a fruit or the like, which has a different color from leaves or branches, in this case, a color television camera captures an image including the object to obtain color image data. This method uses an image processing device to compare the differences in color on the color image data and extract the object.

(Problem to be solved by the invention) However, the objects to be extracted this time are nodes of seedlings such as potatoes, and these nodes have the same color as the leaves and seeds, so the technology to extract the objects based on the difference in color as described above is needed. is not applicable at all.

Also, a node on a seedling is a swollen part of the stem, so when extracting a node on a seedling, you can determine the diameter of the stem and judge it as a node based on its thickness, but the stem diameter increases from the root to the tip. Since it gradually becomes thinner, it is not possible to determine the knot portion based on the thickness using a certain threshold.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a shape recognition device that can reliably extract characteristic parts such as knots even if the color of a plant is the same or the diameter of the stem changes.

[Structure of the Invention] (Means for Solving the Problems) The present invention provides an object to be recognized by irradiating one-dimensional light at a predetermined angle in a direction substantially perpendicular to the direction in which the object to be recognized extends. A light irradiation means for obtaining a light section shape of , an imaging device disposed above a position on the object to be recognized where the light section shape is formed, and a light irradiation means that relatively moves the object to be recognized and the light irradiation means. irradiation position moving means for moving the position of the light section shape formed on the object to be recognized at predetermined intervals in the direction in which the object to be recognized extends, and the light section shape of the image data obtained by imaging with the imaging device. Width detection means for detecting the edge position of the recognized object and determining the width of the recognized object from this edge position, and detecting the characteristic part of the recognized object from the change in the width of the recognized object determined by the width detecting means. This is a shape recognition device that attempts to achieve the above object by including a feature detecting means.

The present invention also provides a light irradiation means for obtaining a light section shape of the recognition object by irradiating one-dimensional light at a predetermined angle in a direction substantially perpendicular to the direction in which the recognition object extends; An imaging device disposed above the position where the light section shape of the object to be recognized is formed, and a light section shape formed on the recognition object by relatively moving the object to be recognized and the light irradiation means. An irradiation position moving means for moving the position at predetermined intervals in the direction in which the object to be recognized extends, and a cross-sectional area corresponding to the thickness of the object to be recognized from the shape of the light section of the image data obtained by imaging with the imaging device. A shape that attempts to achieve the above object by comprising a cross-sectional area calculation means to be sought, and a characteristic part detection means for detecting a characteristic part of the recognized object from a change in the cross-sectional area of the recognized object calculated by the cross-sectional area ffi calculation means. It is a recognition device.

(Function) With the provision of such a means, one-dimensional light is applied to the light irradiation means in a direction substantially perpendicular to the direction in which the object to be recognized extends, and at a predetermined angle with respect to the object to be recognized. The object is irradiated with light to obtain a light section shape of the object to be recognized, and this is imaged by the imaging device. Further, in this state, the object to be recognized and the light irradiation means are relatively moved by the irradiation position moving means, and the position of the light section shape formed on the object to be recognized is moved at predetermined intervals. On the other hand, the width detection means detects the edge position of the object to be recognized from the shape of the optical section of the image data obtained by imaging with the imaging device, and determines the width of the object to be recognized from this edge position. Characteristic parts of the recognized object are detected from changes in the width of the recognized object.

Also, by having the above means, the cross-sectional area calculation means calculates the cross-sectional area according to the thickness of the object to be recognized from the shape of the optical cross-section of the image 1 (P! data) obtained by the imaging device, and The characteristic portion detecting means detects a characteristic portion of the object to be recognized from a change in the cross-sectional area of the object to be recognized.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 is a configuration diagram of a shape recognition device applied to extract nodes, which are one of the characteristic parts of seedlings. A mounting table 2 is provided on the XY table 1, and on this mounting table 2, for example, a potato seedling 3 is placed. Note that the potato seedlings 3 are arranged so that the direction in which they extend coincides with the X direction. On the other hand, a slit light source 4 is arranged above the mounting table 2, and this slit light source 4 emits slit light 5 in the y direction, that is, a direction substantially perpendicular to the direction in which the potato seedlings 3 extend, onto the mounting table. The beam is irradiated at an angle of 45 degrees with respect to the surface. Also, 1 of the slit light 5 at 1113
Imaging device 6 is placed directly above the 1.1η1 point to take an image of +Yi 3 of the potato.

Incidentally, the XY table 1 is configured to move in the X direction at predetermined intervals in response to a drive control signal from the movement control section 7.

The image signal output from the imaging device 6 is transmitted to both image portions f11.
! is being sent to the device 10. This imaging device 10 has A/
A D (analog/digital) converter 11 and an image memory 12 are provided, and the image signal from the imaging device 6 is converted into an A/D converter 11 and an image memory 12.
D converter] is converted into a digital image signal and stored in the image memory 12 as image data. Further, this image processing device 10 is equipped with a main control section 13, and this main control section 13 includes an image memory 12 and a preprocessing section 1.
4. A distance calculation section 15, a peak position detection section 16, a stem direction extraction section 17, a width detection section 18, a cross-sectional area calculation section 19, a surface detection section 20, and an output section 21 are connected. The preprocessing section 14 has a function of performing smoothing, -order differential processing, binarization processing, thinning processing, and smoothing processing on the image data stored in the image memory 12. The distance calculation unit 15 has a function of calculating the height of the stem of the potato seedling 3 based on the shape of the optical section of the preprocessed image data. The peak position detecting unit 16 functions as a stem height position detecting means for detecting a high position portion of the stem, that is, a peak position, from the shape of the optical cross section of the preprocessed image data. The stem direction extraction unit 17 has a function of extracting the direction in which the stem extends by tracing the peak position of the stem detected by the peak position detection unit 16 for each optical section shape obtained each time the X table is moved. It is. Furthermore, the width detection section 18 has a function of determining both edges of the stem from the shape of the optical section of the preprocessed image data, and determining the width of the stem from these edges. This function has the function of determining the cross-sectional area of the stem from the shape of the optical section of the image data. The surface detection unit 20 has a function of detecting a node in a stem from either or both of a change in the width of the stem and a change in the cross-sectional area of the stem. The main control section 13
According to the three-dimensional shape capture flowchart shown in the figure and the surface detection flowchart described later, the preprocessing section 14, distance calculation section 15, peak position detection section 16, stem direction extraction section 17, width detection section 18, cross-sectional area calculation section 19, surface It has a function of issuing commands to the movement control section 7 through the detection section 20 and further through the output section 21.

Next, the operation of the apparatus configured as described above will be explained with reference to the three-dimensional shape capturing flowchart shown in FIG. When the potato seedling 3 is placed on the mounting table 2 and the slit light 5 is irradiated onto the potato seedling 3 from the slit light source 4, this slit light 5 cuts the light along the shape of the potato seedling 3. It is formed into a planar shape. When this optical section shape is imaged by the imaging device 6, the image signal output from the imaging device 6 is sent to the image processing device 10. In step s1, this image processing device 10 manually converts the signal into a digital image signal using the A/D converter 11 and stores it in the image memory 12.
3 (stored as image data. In this image data, g is the part of the light cross section shape where the slit light 5 is irradiated on the seedling 3, and R2 and ρ3 are the parts of the light cross section where the slit light 5 is irradiated. is the part irradiated onto the surface of the mounting table 2.In this image data, the part where the slit light 5 was first irradiated, that is, the boundary line on the right side of the drawing, best expresses the old state of the seedling 3. In the image data, the part irradiated by the slit light 5 is bright and the other parts are dark, but in the image data from FIG. 3 onwards, the part irradiated by the slit light 5 is shown dark for the sake of illustration.

When the image data is thus stored in the image memory 12, the main control section 13 issues an operation command to the preprocessing section 14. This preprocessing unit 14 performs smoothing (moving average) processing and negative-order differential processing on the image data, and then performs binarization processing at a predetermined level. Then, the preprocessing section 14 thins the image data after the binarization process and further performs a smoothing process to obtain the image data shown in FIG. 4.

When this preprocessing is completed, the main control unit 13 performs step S3.
An operation command is issued to the distance calculating section 15 at . Upon receiving this command, the distance calculation unit 15 calculates the distance from the potato seedling 3 to the imaging device 6 by applying the triangulation method using the preprocessed light section shape. The results of determining the height of this seedling 3 are shown in FIG. In this way, three-dimensional information about the shape of the light section at one location in the potato lily 3 can be obtained.

Next, in step S5, the main control unit 13 determines whether the image capture for the entire seedling 3 has been completed, and if it has not been completed, the main control unit 13 proceeds to step S0 and issues a command to the movement control unit 7 to move the X table. As a result, the X table of the XY table 1 moves by a predetermined distance. Hereinafter, three-dimensional information on the shape of the light section at each different position in the seedling 3 is obtained in the same way as the effect described above. In this way, the shape of the optical section at each position of the entire potato seedling 3 as shown in FIG. 6 is obtained.

Next, the main control section 13 issues a command according to the w1 detection flowchart shown in FIG. That is, at step r t +=, the main control section 13 issues an operation command to the peak position detection section 16. In response to this command, the peak position detection unit 16 detects the peak position p at the optical section shape 5(0) of the root of the potato seedling 3 in the image data shown in FIG. seek. In addition,
This peak position po is used to determine the high part of the stem, and does not necessarily have to be the maximum value of the convex cross section.

Peak position po at root optical section shape S (0)
Once determined, the main control section 13 moves from step f'l to step r2 and issues an operation command to the stem direction extraction section 17. This stem direction extraction unit 17 traces each peak position of each optical section shape according to the base peak extraction flowchart shown in FIG.

That is, the position is to be traced as close as possible to the ridge line m of the seedling. Specifically, each light section shape 5 (1
), 5(2)...S (k)...5(n-1). In step S1, k is set to "0", and in the next step S2 at peak position p. The slope of the tangent j at is determined.

Then, it is determined whether this slope is "less than or equal to 0", "equal to O", or "greater than or equal to 0". Here, in the optical section shape 5(0), since the slope of the tangent j is "0 or more", Do is located in the negative direction of y when viewed from the ridge line m of the seedling. Therefore, the process moves from step S5 to g8, and the displacement Δy to be corrected in the y direction is set to "10α". However, in step S9, the peak position pl in the next light section shape 5(1) is "+α" in the y direction when viewed from the light section shape 5(0).
” is determined at the position displaced by ”. Then step glo
, gll, the process returns to step S2 again, and the slope of the tangent at the peak position p+ in the optical section shape 5(1) is determined in the same manner as described above. Then, by determining the slope of the tangent line, the slope is less than or equal to 0, so the peak position p1 is located in the positive direction of y when viewed from the ridge line m of the seedling.Therefore, the process moves from step S3 to g6, and the slope is Displacement Δyl to be corrected
J<r−α”. However, in step S9, the peak position p2 in the next light section shape 5(2) is determined to be a position displaced by "-α" in the y direction when viewed from the light section shape 5(1). Hereinafter, in the same way, each beam position p in each optical section shape 5(2) to S(n) is sequentially
2 to pn are determined, and this operation is executed until there are no more operations. In this way, each peak position p. By determining -p n-1, for example, if the stem is curved to the right in the drawing, the position of the vehicle 1 will always be corrected in the negative y direction. Figure 1O shows the results of detecting each peak position, each peak position is indicated by "・", and by connecting these peak positions, the number of potato seedlings
The stems are extracted.

Next, return to the node detection flowchart shown in FIG.
In this step, the main control section 13 issues an operation command to the width detection section 18 . This width detecting section 18 first detects each peak position p determined by the peak position detecting section 16. - Find both edges of the stem based on pn.

To explain how to find this edge with reference to Figure 11,
The light section shape 5(0) is shifted from the peak position p0 to the left and right at predetermined intervals. Then, it is detected that both shifted points are higher than a predetermined height from the surface 2a of the mounting table 2 and the inclination at that point is a predetermined angle, for example, 80 degrees or more. In this way, both points bl, b
2 is detected, the interval between these points bl and b2 is determined as the width W of the stem. Similarly, each light section shape 5
Edges are determined starting from each of the peak positions p1 to pn of (1) to 5(n), and the width W of the stem is determined from these edges. FIG. 12 is a diagram showing the detection results of the edge positions of the stem, where each edge position is indicated by "." and the width W of the stem is shown in correspondence with the position of the stem.

Further, in step r3, the main control section 13 issues an operation command to one cross-sectional area calculation section. When this cross-sectional area calculation section 1 receives an operation command, it calculates the cross-sectional area of the stem using the following method. That is, as shown in FIG. 13, the optical section shape 5 (n-1)
Each height from the surface 2a of the mounting table 2 is measured at predetermined intervals between both edges L+3°b4 found in . ,hl...h
Find n-1. And each of these heights. +hl・
The cross-sectional area SQ of the stem is calculated by integrating h n-1. That is, SQ-Σhk. Each cross-sectional area SQ obtained in this way is the first
This is shown in Figure 4.

Here, the node on the stem of the potato seedling 3 is a part that is thicker than the diameter of the stem. Therefore, Fig. 12 and 1
In the width W and the cross-sectional area SQ of the stem shown in FIG. 4, the positions where the width W becomes large and the places where the cross-sectional area SQ becomes large are common, and correspond to the node portions of the stem, respectively. However, it is difficult to determine the exact location.

Therefore, in the next step f4, the main control section 13 issues an operation command to the knot detection section 20. This node detection section 20
performs differential processing on the width W and cross-sectional area SQ shown in FIGS. 12 and 14 in the direction in which the stem extends from the root to the tip. As a result of this differentiation process, the width W and the cross-sectional area SQ suddenly change, respectively, at points el and e2.
...if, 12... is extracted. Then, the node detection unit 20 detects these cl, e2...11, 12...
・Extract each part as a clause.

In this way, in the above embodiment, the slit light 5 is irradiated onto the stem to obtain the shape of the light section at each position of the stem, and the edge position of the stem is detected from the shape of the light section. Since the convergence of the stem is determined from the change in convergence, nodes in the stem can be detected from the change in convergence. Nodes in the stem can be reliably extracted even if they are thin and changing.

Further, in the above embodiment, the cross-sectional area of the stem is determined from the shape of the optical section of the image data, and nodes in the stem are detected from changes in these cross-sectional areas.

In this case, the cross-sectional area of the stem also includes information about the depth of the stem, so even if the shape of the node changes in the depth direction, it can be reliably extracted as a node, and it is better than when extracting only by changing the width W. Note that the present invention is not limited to the above embodiment, and may be modified without departing from the spirit thereof. For example, clauses may be extracted by the method shown in FIG.

That is, in the figure, Sg is the shape of the optically cut surface of the stem, and b5. When bfi is the edge position of the stem obtained from this optical section shape Sg by the method shown in FIG. 11, the lens 6a of the imaging device 6 is
A line is drawn perpendicular to the plane E on which .

Then, these lines, the light section shape Sg, and the lens 6a
The area SS surrounded by the plane E is calculated. here,
This area SS becomes smaller as the diameter of the stem becomes larger. However, since the diameter of the stem increases at the node, the node can be extracted by detecting the amount of change in this area SS.

Further, the light irradiated onto the seedlings 3 may be limited to slit light, but the laser light may be expanded into a slit shape using a cylindrical lens, or may be scanned using a rotating mirror or a polygon mirror. Furthermore, when obtaining the shape of the optical cross section at each position, the imaging device 6 and the slit light source 4 may be moved integrally in the X direction instead of moving the XY table 1. Furthermore, when extracting a node, either a change in the width W of the stem or a change in the cross-sectional area SQ of the stem may be performed.

Furthermore, the device of the present invention can be applied not only to the extraction of nodes of seedlings but also to the extraction of branching parts of trees, etc., and the extraction of branch points or defects in electrical circuit networks.

[Effects of the Invention] As described in detail above, according to the present invention, it is possible to provide a shape recognition device that can reliably extract characteristic parts such as nodes even if the color is the same or the diameter of the stem changes.

[Brief explanation of drawings]

1 to 15 are diagrams for explaining an embodiment of a shape recognition device according to the present invention, and FIG. 1 is a configuration diagram;
Figure 2 is a flowchart for capturing three-dimensional shapes, Figure 3 is a schematic diagram of slit image data of seedlings, Figure 4 is a schematic diagram of smoothed image data, and Figure 5 shows the calculation results of stem height. figure,
Figure 6 is a schematic diagram of the optical section shape of the entire bacterium, Figure 'TS7 is a flowchart of node detection, Figure 8 is a flowchart of peak position tracking, Figure 9 is a schematic diagram to explain peak position tracking, 1st
Figure 0 is a diagram showing the tracking results of the peak position, Figure 11 is a diagram to explain the width calculation method, Figure 12 is a diagram showing the extraction of nodes from the width, and Figure 3 is a diagram to explain the method of calculating the width. FIG. 14 is a diagram for explaining the calculation method, FIG. 14 is a diagram showing the operation of extracting nodes from the cross-sectional area, and FIG. 15 is a diagram for explaining another way of extracting nodes. 1...XY table, 2... mounting table, 3... potato fungus, 4... slit light source, 6... imaging device,
7... Movement control unit, 10... Image processing device, 12.
... Image memory, 13... Main control unit, 14... Preprocessing unit, 15... Distance calculation unit, 16... Peak position detection unit, 17... Stem direction extraction unit, 18...・Width detection part,
19... Cross-sectional area calculation unit, 20... Node detection unit.

Claims (2)

[Claims] (1) a light irradiation means for obtaining a light section shape of the object to be recognized by irradiating one-dimensional light at a predetermined angle in a direction substantially perpendicular to the direction in which the manufacturing device extends; The light cut is formed on the recognized object by relatively moving an imaging device disposed above a position in the object where the light cut surface shape is formed, the recognized object and the light irradiation means. irradiation position moving means for moving the position of the surface shape at predetermined intervals in the direction in which the object to be recognized extends; Width detection means for detecting a position and determining the width of the object to be recognized from this edge position, and a feature for detecting a characteristic part of the object from a change in the width of the object to be recognized determined by the width detection means. 1. A shape recognition device comprising: a portion detection means. (2) a light irradiation means for obtaining a light section shape of the object to be recognized by irradiating one-dimensional light at a predetermined angle in a direction substantially perpendicular to the direction in which the object to be recognized; The light is formed on the recognition object by relatively moving an imaging device disposed above a position in the recognition object where the light section shape is formed, and the recognition object and the light irradiation means. irradiation position moving means for moving the position of the cut plane shape at predetermined intervals in the direction in which the object to be recognized extends; A cross-sectional area calculating means for calculating a cross-sectional area according to the thickness, and a characteristic part detecting means for detecting a characteristic part of the recognized object from a change in the cross-sectional area of the recognized object determined by the cross-sectional area calculating means. A shape recognition device characterized by comprising: