JPH0782548B2 - Circle figure center detection method - Google Patents

Description

Detailed Description of the Invention [Object of the Invention]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a circle figure center detection method used to detect the center of a circle figure included in a digital image that has been binarized. (Prior Art) Conventionally, as a method for detecting the center of a circular figure in a binarized digital image, a method for detecting the center of a digital image f (i, j) shown in FIG. Some use the matching filter h (l ₁ , l ₂ ) shown in FIG. The digital image f (i, j) is composed of N × N pixels, contains a circular figure R after input or preprocessing, and noise (not shown). ", And other parts are" 0 ". The matching filter h (l ₁ , l ₂ ) is composed of M × M pixels,
Set "1" to the pixel corresponding to the contour of the circular figure R,
"0" is set in other pixels. A matching filter h (l ₁ , l ₂ ) is superposed on the digital image f (i, j), and the pixel values are multiplied by each other according to the following equations to obtain a filter output g (i, j). Obtainable. To detect the center of the circular figure R, the digital image f
The matching filter h (l ₁ , l ₂ ) is scanned over (i, j), and the filter output g (i, j) is obtained for all the pixels of the digital image f (i, j). Then, the matching filter h that maximizes the filter output g (i, j)
Matching filter h (l ₁ , l ₂ ) when the position of (l ₁ , l ₂ ), that is, the circular figure R and the matching circle match
The center of the circular figure R is detected from the position of. (Problems to be Solved by the Invention) However, in the above-described conventional method for detecting the center of a circle figure, since the matching filter h (l ₁ , l ₂ ) is used, the circle figure R is naturally used. It is necessary to set pixels (set filter coefficients) in advance, and at this time, the filter coefficient must also match the line width of the circular figure R. Therefore, for example, when the radius of a circle changes,
Since it is necessary to use a filter coefficient prepared in advance every time the image changes, there is a problem that the process for detecting the center is extremely complicated. (Object of the Invention) The present invention has been made in view of such a conventional problem, and can eliminate the matching filter and simplify the process for center detection. An object of the present invention is to provide a method for detecting the center of a circular figure that can easily detect the center even when the position or size of the object changes.

[Constitution of the invention]

(Means for Solving the Problems) The method for detecting the center of a circle figure according to the present invention detects the center of a circle figure in a digital image that has been binarized within a symmetrical range about an arbitrary axis. Of the pixel values of the axes are multiplied by the output values of the axes, and two axes having the extreme values of the outputs are detected in different directions. At this time, the function is j on one axis and i on the other side. And the screen size is N,
Let n be the radius of the circular figure, k be the variable in the region of 1 to N with respect to each axial direction, and be 1 be the variable of the region of 1 to n that determines the symmetry range around each axis. It is characterized in that the intersection of both axes where the output becomes an extreme value is detected as the center of the circular figure. (Example) Hereinafter, the present invention will be described with reference to the drawings. 1 to 5 are views for explaining an embodiment of the present invention. That is, the digital image f (i, j) is composed of N × N pixels as shown in FIG. 1, and contains a circular figure R and a noise component (not shown). Partial pixels are "1" and other pixels are "0". In the figure, x and y are digital images f (i,
The horizontal direction of j) and the vertical direction orthogonal thereto are shown. In this embodiment, the axes for center detection correspond to the x and y directions. To detect the center of the circular figure R, first, x,
The function [gx (j), gy (i)] is used as the output of the axis j, i by multiplying the pixel values in the symmetrical range around the axis j, i in the y direction, and here the output gx (j) , gy (i) is defined as follows. Here, the number of pixels, variables, etc. used in the calculation use common signs in the x and y directions, and in the above equation, n is the radius of the circular figure and k is 1 with respect to the directions of the axes j and i. Is a variable in the region of 1 to n, and l is a variable in the region of 1 to n that determines the symmetric range about each axis j, i. That is, the above equation becomes
As shown by the broken line in the drawing, AND operation is performed on all pixels at symmetrical positions in a symmetrical range centered on the axis j (axis i). The maximum range is N (screen size; pixel row) x n (radius of circular figure R)
Is. Next, the axes j, i are scanned in the y, x directions, respectively, and the outputs gx (j), gy (i) of the axes j, i are obtained for the entire screen. 2 to 4 are diagrams for schematically explaining the above-mentioned processing, taking the axis j in the x direction as an example. The axis j
Scan in the y direction from the upper side of the digital image f (i, j). For convenience of explanation, it is assumed that the circular figure R is located substantially at the center of the digital image f (i, j), and the symbols a to c are added to the axis j to indicate the scanning position. First, as shown in FIG. 2A, when the axis ja does not cross the circular figure R and a part of the circular figure R is included below the symmetric range indicated by the broken line, As shown in FIG. 2 (b), if the lower side of the symmetrical range is folded upward, the pixels at symmetrical positions are "0,0" and "0,1". According to the output definition, the output gx (ja) of the axis ja is "0". Next, as shown in FIG. 3 (a), the axis jb
If the circle crosses the circumference of the circle R, the circle R is partially overlapped as shown in FIG. 3 (b) and the circle R partially overlaps with each other. The output gx (jb) for 1 "is obtained. Further, when the axis jc crosses the center of the circular figure R as shown in FIG. 4 (a), the circular figure R is folded in half as shown in FIG. 4 (b). , As a result of maximizing the range where pixels are "1,1", the output gx (jc) becomes the maximum value (extreme value). In other words, the axis jc of the position that gives the maximum output value is a circle It is the axis of symmetry of R. Therefore, by performing the same process for the axis i in the y direction, as shown in FIG. 4 (c), the axis ic that maximizes the output gy (i) can be detected. Axis jc, i
The intersection C of c can be detected as the center of the circular figure R. In practice, since the entire area of the digital image f (i, j) is processed, it is possible to obtain a continuous output line as shown in FIG. In this case, as is clear from the above description, since the circular figure R has a predetermined line width, the output line has a peak at a portion corresponding to the axis of symmetry of the circular figure R and gradually decreases on both sides thereof. Make up. Also, since the digital image f (i, j) contains miscellaneous components,
The output line may show a value in a portion where the circular figure R does not exist, but since there is no large axially symmetric noise component, it does not affect the center detection. FIG. 5 is a flow chart for explaining the process of the above process in the digital computer. First, in step 10, by inputting image data, that is, the binarized digital image f (i, j), in step 11, the screen size (example of pixel) N
And a constant A are set, and the radius of the circular figure on the digital image f (i, j) is set by the constant A. The constant A is determined by, for example, the size of the actual circular figure to be imaged, the characteristics of the imaging device, and the like. Next, in step 12, the maximum values of the outputs in the x and y directions are set to zero as dummy variables x max and y max in the program, and at the same time, the calculation is performed in the x and y directions at the same time. The axis (symmetry axis) ic, jc is set to zero as the axis i. Further, in step 13, the value of the axis i is replaced with i + 1 and the output gx of the axis in each direction is changed.
(I) and gy (i) are set to zero, and the variable k of the screen size area in the direction of the axis is set to zero. Then, in step 14, the variable k is replaced with k = 1 and the axis is set to the center. The variable l for determining the symmetry range to be set is set to zero, and then in step 15, the variable l is replaced with l + 1. Next, in step 16, the value of the axis i, the symmetric range variable l, and the calculated value α based on the function using the screen size N,
β is obtained, and in step 17, the outputs gx (i) and gy (i) of the axes are obtained using the arithmetic units α and β. After this,
In step 18, it is compared whether or not the value of the variable l in the symmetric range is greater than or equal to the value of the radius n of the circular figure. If it is more than (yes), the process proceeds to step 19, and if it is less than (no), Is
Return to step 15 as shown. In step 19, it is compared whether or not the value of the variable k with respect to the direction of the axis is greater than or equal to the value of the screen size N. If it is more than (yes), the process proceeds to step 20, and if it is less than (no), , Return to step 14 as shown. In step 20, it is compared whether or not the value of the output gx (i) of the axis in the x direction is equal to or more than the maximum value x _max of the dummy variable, and if it is (yes), in step 21, the dummy variable Maximum value x _max and shaft output gx
(I) is equal, and the axis i is replaced with the symmetry axis ic in the x direction that was to be obtained, and then the process proceeds to step 22. In the following case (no), the step 21 is omitted and the step is omitted. Proceed to 22. In step 22, it is compared whether or not the value of the y-axis output gy (i) is greater than or equal to the maximum value y _{max of the} dummy variable, and if it is greater than or equal to (yes), step 23
, The maximum value y _max of the dummy variable is equal to the axis output gy (i), and the axis i is replaced with the symmetric axis jc in the y direction that was to be obtained, and then the process proceeds to step 24, and the following (no) In the case of,
Proceed to 24. In step 24, it is compared whether or not the value of the axis i is equal to or more than half the value of the screen size N, and the result is (yes)
In the case of, the process proceeds to step 25 to detect the center coordinates of the circular figure from the intersection of the pair of symmetry axes ic and jc, and in the case of (no) below, the process returns to step 13 as shown in the figure. That is, according to the circle figure center detection method described in the above embodiment, the center of the circle figure R can be reliably detected even if the circle figure R moves or its radius changes. When the radius of the circular figure R is intentionally changed, it is sufficient to change the value n, and therefore the operation is extremely simple. Further, it is of course possible to carry out the calculation over the entire symmetric range centered on an arbitrary axis i, j, but as in the above embodiment, the symmetric range is defined by the variable l having the radius n of the circular figure R as the region. By making the determination, the amount of calculation can be reduced without impairing reliable center detection. (Application Example) FIGS. 6 to 11 are views for explaining an application example of the present invention, showing a case where the present invention is applied to a melting apparatus used for heating and melting a material to be melted in a zero gravity or extremely low gravity environment. Shown in the example. That is, as shown in FIG. 6, the melting device 30 divides the inside of a cylindrical casing 31 by two partition walls 32a, 32b to form three storage chambers 33a, 33b, 33c, and one side ( Sixth
In the accommodation chamber 33a (upper side in the figure), a rate gyro 34 and
A battery 35, a computer 36, and an output distribution device 37 are housed, a furnace body 38 having a substantially spherical shape is housed in an intermediate housing chamber 33b, and a fuel cylinder 40 of a thruster 39 is housed in the other housing chamber 33c. Are housed. The thruster 39 is provided at several places on the side surface of the casing 31, and is provided with a plurality of injection nozzles 39a that are opened in appropriate directions. The furnace body 38 is provided with a plurality of irradiation portions 41 that open toward the center of the furnace body 38 and are projected outward. The irradiation unit 41 has a reflecting surface formed of a parabolic surface inside thereof, and is provided with a lamp such as a halogen lamp or a xenon lamp inside the vertex position, and the heat inside the furnace body 38 is provided outside the vertex position. A heat sink 41a for absorbing and diverging is provided. In order to improve the irradiation efficiency, the irradiation part 41 is provided in an arrangement corresponding to the apex of the regular polyhedron with respect to the inner surface of the furnace body 38, and as shown in FIG. The material M to be melted floating in the interior is irradiated with light intensively to heat and melt the material M to be melted. Further, the furnace body 38 is provided with two sets of cameras 42a, 42b, 43a, 43b for detecting the position of the melted material M. This position detecting means is based on the stereo camera method for detecting the position of the object on the three-dimensional coordinates by the two-dimensional images obtained from different directions, but here, it is opposed to each other on the center line of the furnace body 38. A pair of cameras 42a, 42b (43a, 43b) are set as one set. The cameras 42a, 42b, 43a, 43b are
As shown in the figure, when the irradiation parts 41 are arranged in a regular octahedron, the irradiation parts 41 are set to face each other on a line passing through the furnace body 38 and the centers of the opposing triangles. Also, each camera 42
The a, 42b, 43a, and 43b are provided with filters 44 having different reflectances and shapes from those of the melted object M so that the cameras facing each other are in the field of view to clarify the imaged melted object M. Each of the cameras 42a, 42b, 43a, 43b is connected to the computer 36. Here, the circle figure center detection method according to the present invention is used for signal processing in the computer 36, and inputs a two-dimensional image captured by each of the cameras 42a, 42b, 43a, 43b,
After detecting the center position of the melted object M which is a circular figure based on the previous embodiment, the position on the three-dimensional coordinates is detected. The melting device 30 heats and melts with the melted material M floating in the center of the furnace body 38 while floating in a zero-gravity or extremely low-gravity environment, and during this time, relative displacement between the furnace body 38 and the melted material M. When occurs, the signal for attitude control from the rate gyro 34 and the image signal from each camera 42a, 42b, 43a, 43b are input to the computer 36 to process the displacement in real time,
Further, by calculating a position control amount for following the displacement and sending a command signal to the output distribution device 37, gas is ejected from the selected nozzle 39a of the thruster 39 and moved by its thrust. Therefore, the melting device 30
Moves following the displacement between the furnace body 38 and the melted material M, and operates so as to always keep the melted material M at the center of the furnace body 38. Further, when the inside of the furnace body 38 is captured by a camera, the light of the irradiation unit 41 is strong and affects the image, so that the portion of the irradiation unit 41 substantially becomes a blind spot. However, since the melting device 30 captures one two-dimensional image with the pair of cameras 42a, 42b (43a, 43b), as shown in FIG. 9, the blind spots D should be complemented with each other. However, as a result, the field of view can be expanded and the detection accuracy can be improved. Further, at least four cameras are required for the position detection, but since the computer 36 uses the center detection method of the circular figure, the cameras 42a, 42b, 43 are used.
It is possible to greatly simplify the processing of signals input from a and 43b. In addition, when the arrangement of the irradiation unit 41 is based on a regular octahedron as described above, or based on a regular dodecahedron and a regular icosahedron, as shown in FIGS. Since the triangles or pentagons are displaced from each other about their central axes, the blind spots can be complemented by making a pair of cameras face each other on this central axis. Further, as shown in FIGS. 11A and 11B, the irradiation unit 41
When the arrangement is based on a regular hexahedron, by providing a pair of cameras 42a and 42b on the line connecting the center point of the furnace body 38 and the point separated from the center of one quadrangle by a predetermined distance l. It is possible to prevent the blind spots of the cameras 42a and 42b from overlapping. The value of the distance 1 can be appropriately set depending on the diameter of the furnace body 38 and the size of the irradiation section 41. FIG. 12 is a diagram for explaining another application example of the present invention, and illustrates an example of application to an acceleration measuring device used for measuring a minute acceleration in a zero gravity or extremely low gravity environment. That is, the acceleration measuring device 50 is a container 52 forming a closed space 51, a target 53 floating in the closed space 51, an image processing device 54 for detecting the displacement of the target 53, and the image processing. A differential operation device 55 for second-order differentiating an output signal from the device 54 and outputting it, and a pair of arms 56, 56 for returning the target object 53 to the measurement start position are provided. The container 52 is a pair of cup-shaped container members 52a, 52b.
To form a closed space 51, and the inside thereof is kept in a vacuum state (for example, about 10 ⁻⁷ Torr). The target 53 has a spherical shape most suitable for three-dimensional displacement measurement, and is covered with a phosphor 53a made of, for example, calcium sulfide or barium sulfide. The image processing device 54 includes two two-dimensional image processors 57a, 57.
It is provided with b and a three-dimensional coordinate calculator 58. The two-dimensional image processor 57a, 57b is equipped with an image input unit 59a, 59b using a TV camera system or fiber system,
The image input sections 59a and 59b are attached to the wall of the container 52.
At this time, one of the image input units 59a displays the x in the closed space 51.
The two-dimensional image based on the axes and the y-axis is located, and the other image input unit 59b is located in the position to capture the two-dimensional image based on the y-axis and the z-axis in the closed space 51. Here, the method of detecting the center of a circular figure according to the present invention, the two-dimensional image processor 57a,
It is used for signal processing in 57b, and detects the center position of the target object 53 which is a circular figure on each two-dimensional coordinate. Further, the three-dimensional coordinate calculator 58 is a two-dimensional image processor 57a, 57
Based on the data from b, the displacement x (t), y (t), z (t) of the target 53 on the three-dimensional coordinates is detected, and this is sent to the differential operation device 55 as data. The differential operation device 55 is configured by connecting two differentiators 60a and 60b, and the displacement x (t), y (t), z (t) of the target object 53 input from the image processing device 54. The second-order factorization of time gives the acceleration (t) in the x, y, and z-axis directions,
Calculate (t) and (t). The differential calculation device 55 includes a display device (including a recorder) 61 for displaying the calculated acceleration.
And the drive device 62 for the arms 56, 56 are connected. The arm 56 includes a rod 63 slidably inserted from opposite ends of the container 52, and forms a cup-shaped holding portion 63a at the tip of the rod 63 in the closed space 51. Around the opening of the sandwiching portion 63a, a large number of spokes 64 are provided that are radially arranged around the axis of the rod 63 as a whole, and an illuminating lamp 65 is provided at the bottom of the sandwiching portion 63a. The spokes 64, 64 are in a state of being inclined with their tips approaching the opposing arm 56 side, and when both arms 56, 56 approach each other, they may stagger alternately with a clearance smaller than the diameter of the target 53. It has become. The spokes 64 where the image input units 59a and 59b are located
Is a notch or a transparent material. The acceleration measuring device 50 described above is mounted on a navigation body that orbits the earth in the upper atmosphere, for example, and performs measurement. That is, the acceleration measuring device 50 can be regarded as a zero gravity state due to the balance between the centrifugal force for navigating the orbit and the attractive force of the earth, and in particular, the target object 53 is an extremely dilute gas in the container 52. Since the relative velocity of is small enough to be ignored, the container 52 is in a state of performing uniform velocity motion under zero gravity. However, the navigation vehicle decelerates due to aerodynamic resistance caused by the surrounding rare atmosphere.
Therefore, since the above-mentioned deceleration also acts on the acceleration measuring device 50, the target object 53 maintaining a constant velocity motion is displaced with respect to the container 52, and the acceleration can be measured from the displacement at this time. it can. In this way, when operating by decelerating the navigation body, the acceleration is displayed by the display device 61.
Is displayed as-(t),-(t),-(t). In addition, when performing the above measurement, first,
Is turned on, and both arms 56, 56 are moved closer to each other, so that the target object 53 is moved on the axis of both rods 63, 63 and held by both holding portions 63a, 63a. At this time, the target object 53 is located substantially in the center of the closed space 51. Then
When the arms 56 and 56 move backward, the lights go out after a predetermined time has passed,
The measurement is performed with the target object 53 in a light emitting state. This measurement can be performed until the target object 53 comes into contact with the container 52 or the arm 56, and thereafter, the differential operation device 55 sends a reset signal to the drive device 62 to drive the arm 56 and the illumination lamp 65. The measurement can be restarted by automatically flashing. Since the acceleration measuring device 50 uses the method of detecting the center of the circular figure in the image processing device 54, the signal processing can be simplified without any loss of measurement accuracy. Also,
Since the acceleration measuring device 50 uses the target object 53 coated with the phosphor 53a, compared to, for example, a case where the inside of the container is measured while being illuminated, there is no reflection or the like due to the inner surface of the container, and the captured image is significantly clearer. Therefore, erroneous detection can be prevented and power consumption can be saved. The illuminating lamp may be provided, for example, in the container 52, but if it is provided in the arm 56 as described above, it is possible to irradiate the target object 53 at a distance or to improve the light emitting action. It is even more effective by saving electricity and electricity.

【The invention's effect】

As described above, according to the method for detecting the center of a circular figure of the present invention, the matching filter used conventionally can be eliminated, and the processing for detecting the center can be remarkably simplified. Even if the position or size of the object changes, it has an excellent effect that the center can be detected easily and surely, simplifying the artificial operation related to the processing and reducing the load on the processing device. It is possible to realize high-speed processing and the like.

[Brief description of drawings]

FIG. 1 is an explanatory diagram showing a digital image and axis output in one embodiment of the present invention, and FIGS. 2 to 4 are explanatory diagrams each schematically showing a processing process performed together with axis scanning.
FIG. 6 is a flow chart for explaining the processing steps in a digital computer, FIGS. 6 to 11 show an application example of the present invention, and FIG. 6 is a perspective view for partially explaining the melting apparatus and FIG. Is an explanatory view showing a state of heating and melting by the irradiation unit, FIG. 8 is a perspective view for explaining the arrangement of the camera, and FIG.
The figure is a perspective view for explaining a blind spot portion with respect to the camera, FIGS. 10 (a) and (b) are perspective views of a regular dodecahedron and a regular icosahedron for explaining the arrangement of the irradiation unit, and FIGS. 11 (a) and (b) are Explanatory drawing which shows the position and axis of a camera when arrangement | positioning of an irradiation part is based on a regular hexahedron, FIG. 12 is explanatory drawing which shows the acceleration measuring apparatus which is another application example of this invention with the cross section of a container schematically, First
13 (a) and 13 (b) are explanatory views showing a conventional digital image and matching filter, respectively. f (i, j) ... Digital image, R ... Circle figure, i, j ... Axis, gx (i), gy (j) ... Output (function of axis).

Claims (1)

[Claims] 1. When detecting the center of a circular figure in a binarized digital image, a pixel value within a symmetrical range about an arbitrary axis is multiplied and the function is output as the axis. Two axes having the extreme values of the output are detected in different directions. At this time, the function has j as one axis and i as the other axis, the screen size is N, the radius of the circle is n, and each axis is Let k be a variable in the region of 1 to N with respect to the direction, and l be a variable of the region of 1 to n that determines the symmetric range around each axis. The method for detecting the center of a circle figure is characterized by detecting the intersection of both axes, where the output becomes an extreme value, as the center of the circle figure.