JPS59231407A - Detecting method of central position of tubular body - Google Patents

Abstract

PURPOSE:To enable the detection of the central position with higher accuracy than unit elements, by calculating the degree of matching through template matching, by computing the position of the center of gravity with reference to the degree of matching, and by determining the position of the center of gravity thus computed as the central position of a steel-plate coil. CONSTITUTION:An apparatus employed comprises an ITV camera 1 which is an input device of an image, and a controller element 3 having a high-speed image processing device 2 and a microcomputer. The high-speed image processing device 2 performs a spatial differentiation operation, a symmetry operation, and an image operation including template matching, while the controller element 3 performs the control of the procedure of recognition, the determination of various parameters and a threshold value, the decision of a proposed central point of a coil, the formation of the data on a circle of a template, etc. A digital image is obtained from the ITV camera, a differential operator is made to operate for each picture element of the image, and thereby an edge map is obtained. The template matching of the circle is conducted with respect to the edge map by using a symmetry operator and a symmetry estimation function. A point at which the degree of matching is above a threshold level and at the maximum is determined to be the central point of the inside diameter of the steel-plate coil.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a center position detection method for detecting the center position of a cylindrical object using an image processing technique using an ITV camera, and in particular to a method for detecting the center position of a cylindrical steel plate coil. Regarding the method.

Generally, when a cylindrical object such as a steel plate coil is suspended by an automated overhead crane and transported from the loading position of the steel plate coil to another position, the center of the steel plate coil is It is necessary to accurately detect the position and correctly position the overhead crane relative to the steel plate coil.

Conventionally, a method of processing a video signal from an ITV camera has been proposed as a method of detecting the center position of a cylindrical object such as a steel plate coil. This method is useful when there is a strong contrast between the steel plate coil and the background.

The center position of a steel plate coil can be detected relatively easily by utilizing the line symmetry of a straight line passing through the center of the steel plate coil (see Japanese Patent Laid-Open No. 82967/1983). However, when this method is actually applied to a steel plate coil yard in a steel mill, it is often impossible to accurately detect the center position of the steel plate coil. This is because, in addition to the image of the steel plate coil that is the object, one other image is inevitably mixed into the two-image signal, and the contrast between the object and the background is also small.

Furthermore, the present inventors have proposed a method for detecting the center position of a cylindrical object such as a steel plate coil, which can withstand practical use.

We proposed a method in which several candidate points for the center position are found, the degree of matching at each candidate point and its neighboring points is calculated using template matching, and the point with the maximum degree of matching is determined as the center position (Sumitomo Heavy machinery technical report vo
1.301980 No. 90 pp 52-57). However, the center position and center point determined by this method have accuracy only in pixel units. Therefore, the diameter of the steel plate coil is large.

Moreover, when the distance between the ITV camera and the steel plate coil is large, that is, when the area of the steel plate coil corresponding to a unit pixel is large, the center position of the steel plate coil cannot be determined with sufficient accuracy.

An object of the present invention is to provide a highly reliable method for detecting the center position of a cylindrical object, which is capable of detecting the center position with high precision even in unit pixels.

According to the present invention, a video signal obtained by imaging a cylindrical object such as a steel plate coil is processed.

In a method for detecting the center position of a cylindrical object that detects the center position of a cylindrical object, a calculation is performed on a video signal using a predetermined symmetry operator and a symmetry evaluation function,
Find a primary candidate point related to the center position, calculate the degree of matching at the primary candidate point and a predetermined number of points in its vicinity by tensorate matching, and select the point with the maximum matching degree from the calculation results as a secondary candidate point. Established as a candidate point.

A centroid position is calculated with reference to the degree of matching at the secondary candidate point and a predetermined number of points in its vicinity.

A method for detecting the center position of a cylindrical object is obtained in which the calculated center of gravity position is determined as the center position of the steel plate coil.

This will be explained below with reference to one drawing.

Referring to FIG. 1, a method for detecting the center position of a steel plate coil according to an embodiment of the present invention is shown in the form of a flowchart. Hereinafter, the process will be explained in sequence according to the steps shown in the flowchart.

(1) Input of Dinotal image Quantizes the video signal obtained from the ITV camera.

A dinotal image (hereinafter referred to as 0 image) is obtained. 0 Image size is 120 pixels vertically and 128 pixels horizontally, brightness is 4 bits,
It is l-.

The intensity of the (i, j) pixel of the 0 image is represented by G(i, j). Here, 0≦G(i,j)<15 0<1<127. 0
The radius R of the inner diameter of the coil on the zero image is calculated from the amount of displacement in the vertical direction with respect to the center of the lens of the TV camera.

(3) Creation of edge map A normal differential operator defined by linear equation (1) is applied to each pixel of the zero image.

grad(e)=l a+b+c-g-h-4l+lc
+f+i-a-d-gl ---(1) Here *
a+b*e*...i is the intensity of each pixel in a 3×3 area centered on pixel e, and its arrangement is shown in FIG. Using the value of this differential operator, an edge map (referred to as 8 images) defined below is created.

8 images-(E(i+j) IE(i r j)-1
if grad(i,j)>d;E(i,j)=Oif
grad (i, j) (d) where d is an integer set as a threshold and gr
The definition of ad(t,j) is F), 1<i<126,
1<j≦118.

(4) Extraction of the center candidate point using the symmetry angle The inner diameter of the coil has a circular shape in the E image.

The center point of this circle is obtained at the intersection of two symmetrical center lines. In this system, the symmetry operator and symmetry evaluation function defined below are used to calculate the 1-element (E(i
, j) = 1, a straight line that is the center line of line symmetry of the figure is extracted in two directions (horizontal and vertical). The intersection of these straight lines becomes a candidate point for the center point of the inner diameter of the target coil.

(a) Symmetry operator 8 The symmetry operator that determines the symmetry in the i direction and j direction with the image point ('0' jo) as the center is W (positive integer)
Define the width as follows.

ψ(W; 1o, Jo)=O1fN1o2φψ(Wjo+
jo) = 1 4f N1o4φ here N1o = (n11<n(w, E(io-n, jo)△
E(io+n,jo)=1)ψ(W;lo,Jo)2φ
1fNjo=φψ(W:j,)+j. )=1 if
Njo(φHere △ is logical product or (i, D Kano((1
, Jll<1 (126°1<j<118), then E(
1, J)=o.

(b) Symmetry evaluation function Evaluate whether any straight line i−+r j””J in the E image can be the center line of line symmetry of a figure composed of 1-element in the E image. It is something.

For a certain threshold value S, Φj1j2('':10)≧S
- Straight line 1-1o, j=j such that F l , 12 (W;jo)≧S.

is determined to be the center line of line symmetry.

On the other hand, according to the properties of Φ and F, if one point (1o* J o ) is the center point of the circle, then Fj1J2(V/;lo)'+
'1112 ("yJO)" is.

A local unimodal peak point is reached in the vicinity of l and J, respectively. TS' ” 'C1oV'o2〕'C'o
VJo2] as a suitable neighborhood of 1°IJO.

For all 1s belonging to ['o1”02), ΦJ,j
2(w; i) is upwardly convex, unimodal, and Φj1J2(
w:1o)〉Φj1j2(w;i) [For all j belonging to Jol "o2-', F
i, 12 (W; J) is upwardly convex and unimodal, and Ft 1i
2(w; j.)≧y, 12(w; j) holds.

In this system, first, the radius R6 of the inner diameter of the coil in the G image is
, 1), determine the width W of the symmetry operator and the threshold S of the evaluation function. Next, extract the center line of line symmetry as follows.

SV = (i=iolrijl j 2 (w:io
)≧S and Φj1 j2(w;i) is the local peak point) 5H=(j=jolFt1□2(w;j)≧S and'
+112 (W: jO) is the local peak point) The candidate point for the center of the coil inner diameter is obtained as the intersection of the straight lines belonging to SV and SH (5) Center obtained in the stage before circular template matching Set the candidate points to (ip, i,) (p=
1°..., m, q=1..., n). Each point (
Generate a circle with radius R6 centered at lp・r q ),
A template matching function is defined on 8 images as follows.

■( + p − ] q ) −N(A) Here N(
A) represents the number of elements in set A.

A= ((i, DI(i = )2+(j-jp)2=
R,' is satisfied.

and E(i,j)-1) The center of the inner diameter of the target coil is determined as the point where ■(ip, jp) is maximum. However, in actual processing, the following considerations are taken to improve the reliability and accuracy of identifying the center point of the coil inner diameter.

(a) The radius of the circle for matching is 'RG-4'
Three types of RGI and RGel are used. Target coil and ITV
There is a possibility that the distance from the center of the camera lens may vary, or the radius of the inner diameter of the coil may deviate slightly from RG.

(b) The matching operation is performed using the above-mentioned first center candidate point (i
p , r q ) for each + ( 1 p r
1 q) Perform this on a total of 9 points, including itself and its 8 neighbors, and use them as secondary center candidate points.

(c) Next, extract points from the center candidate point and the eight points in its vicinity where each matching frequency is equal to or higher than a specified frequency (that is, a threshold level), and find their centroid positions and use them as the center position.

This makes it possible to detect the center position within an accuracy of the upper pixel.

The center of gravity position (ig, jg) is determined as follows. The coordinates of the point above the threshold level are (il, jl)・(
12, J2) ...... (jN, jN), and if each matching frequency is P 1J P 2...PN, then the threshold level or higher is shown in Figure 4 below (Fig. Here, we will show an example in which three points (15 or higher) were obtained and their center of gravity positions were determined.

It is assumed that the arrow on the vertical axis and the number above it (inside <>) indicate the matching frequency. Also, the coordinates of the center candidate point are (81, 66)
It is said that

(11・j+) −(80,67), P1= 2
0(i2.j2) = (8L66), P2= 3
6(i5.js) −(8L67), p3=35
Therefore, the center point is (i,, jg) - (80, 8, 66,').

(Points marked with X in FIG. 3) In the following margin, the hardware configuration of a visual sensor that performs the work procedure shown will be described below based on FIG. 4.

This configuration includes an ITV camera 1 as an image input device, a high-speed image processing device 2, and a controller section 3 having a microcomputer. Among these, the high-speed image processing device 2 is in charge of image calculations such as spatial differential calculation, symmetry calculation, template matching, etc., and the controller unit 3 is responsible for controlling the recognition procedure, determining various parameters and threshold values, and determining coil center candidates. Responsible for determining points and generating template circle data.

According to the invention, obtaining a digital image from an ITV camera,
A differential operator is applied to each pixel of the image to obtain an edge map, and circular template matching is performed on the edge map using a symmetry operator and a symmetry evaluation function, and the matching frequency is set to a threshold level. Since the above point and the point of maximum frequency are set as the center point of the inner diameter of the steel plate coil, various other images of the object appear on the image, and the reliability of the image is improved even when there is a large contrast difference between the object and the background. It has the effect of increasing sex.

In the above description, a steel plate coil having a circular shape has been described, but the present invention is not limited to this in any way, and can also be applied to a rectangular shaped coil. Furthermore, the present invention is also applicable to coils of steel plate rigs.

[Brief explanation of drawings]

Figure 1 shows the flow of the procedure for recognizing the center of a steel plate coil. FIG. 2 is a pixel differential operator, FIG. 3 is a graph showing the relationship between matching frequency and power coordinates, and FIG. 4 is a diagram showing the hardware configuration of the visual sensor. 1... High-speed image processing device for ITV Canora and 2. 3...Controller part

Claims (1)

[Claims] 1. In a method for detecting the center position of a cylindrical object, in which a video signal obtained by imaging the cylindrical object is processed to detect the center position of the cylindrical object, a predetermined symmetry is detected with respect to the video signal. A calculation using a degree operator and a symmetry degree evaluation function is performed to obtain a first-order candidate point related to the center position, and the degree of matching at the first-order candidate point and a predetermined number of points in its vicinity is calculated by template matching hx
The point with the highest degree of matching is determined from the calculation results as a secondary candidate point, and the position of the center of gravity is calculated by referring to the degree of matching at the secondary candidate point and a predetermined number of points in its vicinity. A method for detecting the center position of a cylindrical object, characterized in that the calculated center of gravity position is determined as the center position of the cylindrical object.