JPS63217480A - Shape measuring method for object having similar circular sections - Google Patents

Abstract

PURPOSE:To facilitate the measurement of a shape by defining an approximate circle having many contour points as a precisely measured circle or an approximate circle having a small minimum square error as a precisely measured circle after calculating the minimum square error among a registered approximate circle, a final approximate circle and contour points forming said two circles. CONSTITUTION:In case coverage factor between a 3rd approximate circle 81 and a 4th approximate circle 82 is smaller than a prescribed value, it is difficult to decide one of both circles 81 and 82 that can be used as a precisely measured circle. Thus a 3rd measurement start point is obtained for production of a 5th approximate circle. Then the operations are repeated to select two approximate circles where the coverage factor 83 set among circles 81 and 82 and a 5th approximate circle is larger than the prescribed value. If such two approximate circles are not obtained yet, these pieces of information are deleted out of registered data as noise information. When said two approximate circles are obtained, the number of contour points forming both circles are compared with each other. Then one of both circles that has the larger number of contour points 84 is used as a precisely measured circle.

Description

[Detailed Description of the Invention] [Industrial Field of Application] The present invention is for rapidly measuring the cross-sectional shape distribution of a collection of objects having a certain cross-sectional area and a fairly wide distribution of circular similar cross-sections. The present invention relates to a method for measuring the cross-sectional shape distribution of an object having a similar circular cross-section.

[Conventional technology]

It is extremely important to know the state of aggregation of aggregates having a cross-sectional shape distribution of an object having a certain shape distribution, and there is a strong desire for the development of rapid analysis methods.

For example, rubber particles distributed in the form of fine particles in an impact-resistant resin such as ABS can have completely different impact resistance properties depending on the size of the particles and the state of the particle size distribution. It is known that it is extremely important to know the particle size distribution. In addition, it is known that the properties of synthetic fibers produced at high speeds vary significantly depending on their cross-sectional shape, and in order to create fibers with a uniform cross-sectional shape, information about the cross-sectional shape can be quickly obtained. It is desired to develop an analysis method.

Conventionally, information regarding these cross-sectional shapes has been obtained by converting these aggregated objects into electron micrographs and visually measuring the shape and size distribution of the imaged circular-like objects. In the technical field described above, the number of individual circular objects included in one image is large, and the number of samples that need to be measured ranges from hundreds to thousands, which not only takes a long time to measure, but also requires visual inspection. the health condition of the person being judged;
There was a major problem in that measurement data was blurred due to visual fatigue and other factors.

Therefore, methods are being considered to use computers to determine the shape of an aggregate with a similar circular cross section.
For example, a distance conversion method, a method using extracted contour curves, a curve detection method applying the Hough transform method, etc. are known. The distance conversion method is a method of converting a binary image signal into an image with shading at a distance from the contour, detecting a peak formed near the center of a circular object, and finding the center of the circular object and its radius. . The method used is to calculate the curvature of each contour of the axial point sequence extracted from the binary image, and also use the unevenness information of the contour point sequence to find the center of the circle and its radius. It is. In addition, the Rough conversion method determines the type of line you want to detect, converts its shape into an equation,
The contour points in the image are mapped to the parameter space, this process is performed, the coefficients of the detected equations appear as peaks on the parameter space, and this peak is detected to simultaneously determine the center of the circle and its radius. .

[Problem that the invention seeks to solve]

Among the conventional techniques described above, the method using the distance conversion method causes a large error in measuring the center and radius of the circle when there is an isolated point in a circular object as shown in (21) in Figure 2. On the other hand, isolated points as shown in (η) in Figure 2 are important points, and if a processing method is used to treat these isolated points as noise, the center point of the desired circle and its radius This causes a large error in the measurement.

The method using the curvature of the extracted contour is to calculate the curvature of each point from the contour point sequence (31) as shown in Figure 3, and also use the unevenness information to find its center and radius. ,
If noise such as whiskers is present, the curvature profile will contain a lot of noise, making it extremely difficult to process and make accurate measurement difficult.

Since the Hough conversion method is limited in the types of lines to be detected, it is very difficult to measure when there are many types of circular objects that make up a collection of circular objects to be measured. becomes.

[Purpose of the invention]

The present invention has been made in view of the above-mentioned drawbacks, and its purpose is to obtain information regarding the circle of the approximately circular cross section of an unknown object having a circularly similar cross section or an aggregate thereof, such as the size of each individual object, It is an object of the present invention to provide a measuring method for accurately, quickly, and easily measuring the radius or, in the case of an aggregate, the diameter distribution situation,
The gist is as follows: (A) An image conversion means for converting a cross-sectional plane image of an object having a similar cross-sectional shape to a circular shape into image information.(B) Based on the gray-scale image information obtained by the image information conversion means, a circular Means for determining a first measurement starting point in a similar cross section (C) Edge detection is performed radially from the first measurement starting point toward an edge portion of the grayscale image information, and the edge detection portion is set as the first measurement starting point. Means (D) for obtaining contour point group information, determining a first approximate circle center and a first approximate circle radius based on the first contour point group information, and using these to create a first approximate circle;
The edge portion of the grayscale image is detected radially from the center of the first approximate circle toward the edge portion of the grayscale image information, and the edge portion is detected as a second contour point group. (E) means for determining a second approximate circle center and a second approximate circle radius based on the second approximate circle and creating a second approximate circle; Divide into point groups and select the second point from each group.
Select the contour points closest to the approximate circumference of Means (F) for creating an approximate circle of No. 3 and registering its data; Means for determining a second measurement starting point within a circular similar cross section based on the grayscale image information obtained by the image information conversion means of (A); G) Based on the second measurement starting point and the grayscale image information (
C) means, (D) means, and (E) means for creating a fourth approximate circle by the same means as the means (H) collation of the first approximation circle and the fourth approximate circle. If the pupil is not equal to or higher than the predetermined rate, register the 44th approximation center and repeat step (G) to create the 5th approximation circle, and continue this process until the 5th billion yen appears. The method of repeatedly creating the final approximate circle is to compare the final approximate circle with the registered approximate circle whose ratio is above a certain level (Kusaki rate) by the means of (I) (G) or (H), and calculate these approximate circles. Compare the number of contour points that make up the circles, and select the approximate circle with a large number of contour points as the precision measurement circle, or calculate the least square error between the registered approximate circle, the final approximate circle, and the contour points that make up these approximate circles. (J) means for outputting information about the circle based on the precision measurement circle; shape measurement of an object having a circular similar cross section; It's in the method.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a schematic process flow diagram showing an example of the method for measuring the shape of a circular analog according to the present invention, and the process will be explained in accordance with this process flow diagram with reference to each figure.

The sample (11) made of an object with a similar circular cross section in Fig. 1 has a similar circular cross section (12) and a bottom part @3). Good too.

In FIG. 1, a sample (11) is imaged by an imager (14), then converted into an electrical signal and input as a grayscale image (means A).

As shown in FIG. 4, in order to determine the measurement starting point for measuring the shape of a circular object on the gray scale image plane, the latitude (42) is set so that two or more grid points are formed on the gray scale image plane (41). Sutra (
43) and its grid point (44)I (44'
)t (44')... are respectively the first measurement starting point,
Eight means are taken to determine the second measurement start point and the third measurement start point. By providing two or more grid points in the shape measurement image, it is possible to more accurately determine the shape of the circular object obtained using these grid points as measurement starting points.

FIG. 5 shows a method for measuring the shape of a grayscale image of a circular object. First, from the first grid point (51),
The strings (S2) t(52'), (52'), which detect the edge parts radially toward the edge parts of the grayscale image information.
... to perform edge detection and form contour point group information (53). Then each string (52).

Perpendicular bisector of (52') (54), (54')
, (54' doors...), set the intersection of these perpendicular bisectors as the center (55) of the first approximation circle, calculate its radius from the average value of the perpendicular bisectors of each chord, and Approximate circle of 1 (56
) to make (C) perform the means.

As shown in FIG. 6, the center (5@) in FIG. 5 obtained by means (C) is set as the center (61) in FIG. edge detection is performed radially (62), (62')
63), and performs the means (D) of creating a second approximate circle (64) using the average value of the radial radii as the radius of the second approximate circle.

As shown in FIG. 7, three or more groups (73 ), (73'
), (73'), (73''')..., and the second approximate circle circumference (72
) is set as the third contour point group, and after finding the center and radius of the third approximate circle from the third contour point group,
Means (E) for registering the approximate circle as (81) in Fig. 8
The process of making it happen.

Next, from the second measurement starting point for circular shape measurement (
(C) means, (D) means, and (E) means, the fourth approximation circle (82) in FIG. 8 is performed, and (G) means is performed.)
, the means (G) for comparing the coverage ratio (83) between the third approximate circle and the fourth approximate circle registered as shown in FIG. 8 is performed. This coverage can be determined depending on the object to be measured, but
It is usually preferable to set it to about 80% or more.

(G) If the coverage ratio of the third approximate circle and the fourth approximate circle determined by the means is lower than the predetermined value, it is difficult to determine which approximate circle can be used as the precision measurement circle.
Find the measurement starting point of (C) means, (D) means and (E)
Repeat the procedure to create a fifth approximate circle, a third approximate circle,
H) means is performed by repeating the above operation so as to select two approximate circles in which the coverage ratio of the fourth approximate circle and the fifth approximate circle is greater than a predetermined value. If it is not possible to search for two approximate circles whose coverage is greater than a predetermined value even with this operation,
These pieces of information are deleted from the registered data as noise information.

When two approximate circles with coverage ratios equal to or higher than a predetermined amount are found as described above, the number of contour points constituting the nearest million circles is compared, and the approximate circle with the largest number of contour points (84) is used for precision measurement in the present invention. It is used as a circle.

If the two approximate circles compared have the same number of contour points, one of the approximate circles is used as the accurately measured circle used in the present invention.

According to the method detailed above, information regarding the shape of an object having a similar circular cross section can be calculated relatively accurately and quickly. Because it is possible to accurately grasp the analysis of each circular shape and aggregate state analysis information, it is possible to, for example, measure the particle size distribution of latex particles and analyze the causes of changes in yarn properties due to cross-sectional changes due to changes in the synthetic fiber manufacturing process. etc. can be done in a short time.

[Brief explanation of the drawing]

Fig. 1 is a schematic flow diagram showing an embodiment of the method for measuring the shape of an object having a similar circular cross section according to the present invention, Fig. 2 is a binary image of the object to be measured, and Fig. 3 is a binary image of the object to be measured. A diagram showing contour points extracted from the image, Figure 4 is a diagram of grid point creation on a gray scale screen,
Figure 5 is the first approximate circle diagram, Figure 6 is the second approximate circle diagram, Figure 7 is the third approximate circle diagram, and Figure 8 is the coverage rate of the two extracted approximate circles. FIG. Tame 2 flash 32 Exhibition 4 figure 5 (21 Tail 7 figure L4 figure I Tail 8 figure

Claims (1)

Scope of Claims: (A) Image conversion means for converting a cross-sectional plane image of an object having a cross-sectional shape similar to a circle into image information (B) Based on the gray-scale image information obtained by the image information conversion means, Means for determining a first measurement start point within a cross section (C) Edge detection is performed radially from the first measurement start point toward an edge portion of the gray scale image information, and the edge detection portion is detected as a first contour. Means (D) for obtaining point cloud information, determining a first approximate circle center and a first approximate circle radius based on the first contour point group information, and using these to create a first approximate circle;
The edge portion of the grayscale image is detected radially from the center of the first approximate circle toward the edge portion of the grayscale image information, and the edge portion is detected as a second contour point group. (E) means for determining a second approximate circle center and a second approximate circle radius based on the second approximate circle and creating a second approximate circle; Divide into point groups and select the second point from each group.
Select the contour points closest to the approximate circumference of Means (F) for creating an approximate circle of No. 3 and registering the data; and means (F) for determining a second measurement starting point within a circular similar cross section based on the gray scale image information obtained by the image information converting means of (A). G) Based on the second measurement starting point and the grayscale image information (
C) Means, (D) Means for creating a fourth approximate circle by the same means as (E) means (H) Collating the third approximate circle and the fourth approximate circle, and covering the third approximate circle. If the rate is not higher than the predetermined rate, it is registered as fourth approximate circle data, and step (G) is repeated to create a fifth approximate circle, and the fifth approximate circle, third approximate circle, and fourth approximate circle are By comparing the approximate circles with each other and repeating this process until an approximate circle whose coverage reaches a predetermined rate appears, the coverage is greater than a predetermined value by means (I), (G), or (H) of creating a final approximate circle. Compare the registered approximate circle with the final approximate circle, compare the number of contour points that make up these approximate circles, and select the approximate circle with a large number of contour points as the precision measurement circle, or compare the registered approximate circle with the final approximate circle. Means for calculating the least square error between approximate circles and the contour points constituting these approximate circles, and determining the approximate circle with the smaller least square error as a precision measurement circle (J) Outputting information about the circle based on the precision measurement circle 1. A method for measuring the shape of an object having a similar circular cross section, comprising: