JPH0451874B2 - - Google Patents

Description

[Detailed description of the invention]

[Industrial Field of Application] The present invention relates to a method of separating two straight lines at the corners of a figure in an image obtained by an image sensor or the like and individually extracting information on these straight lines. [Prior art] Various assembly operations on a production line, for example,
When automating the assembly of various parts to the body of a car, the body of a home appliance, or an IC board, etc., it is necessary to
Automatic recognition is performed for substrates, etc., but the recognition target usually has a substantially rectangular shape, or at least a part of its outer shape has a right angle that can be used as a reference for recognition. It has portions. Furthermore, even if the recognition target itself is not rectangular, it is often moved on a rectangular trolley or has a rectangular portion on the target. Therefore, by using a substantially rectangular part, more specifically, a right-angled part forming a part of the shape, as a reference, the positional deviation of the object with respect to the reference position and the inclination of the attitude angle with respect to the reference posture are detected. ,
Objects can be easily recognized. Therefore, the present inventors separately proposed a method for measuring the positional deviation and posture angle of a rectangular object by observing the right angle part of the object using an image sensor and finding the equations of two mutually orthogonal lines. There is. In this case, it is necessary to separate and extract the information of the two straight lines at the corners of the figure in the image of a single image sensor using the simplest possible processing, but existing techniques cannot always perform satisfactory separation and extraction. Can not. [Problems to be Solved by the Invention] The present invention separates two straight lines at the corners of a figure in a two-dimensional image of an object obtained by an ITV camera or other image sensor by a simple means,
This provides a method for individually extracting information about those straight lines. [Means for solving the problems] In order to achieve the above object, in the present invention,
After the image of the corner of the figure formed by the intersection of two straight lines obtained by an image sensor or the like is completely binarized, a spatial filter is used to calculate the real part of the figure on either side of the n×n micro matrix. For each combination of direction codes that contrast with the attitude angle of the straight line part at the corner of the figure, add the total number of direction codes, and then find the set of direction codes that shows the maximum value. A technical means of extracting combined components, separating direction codes corresponding to the two straight lines from among the combined components, and calculating a linear equation of the two straight lines based on data corresponding to each separated direction code. is adopted. [Example] The method of the present invention will be explained in more detail below with reference to the drawings. As shown in FIG. 1, it is assumed that there is a reference position of the object indicated as A ₀ in the space where the object A whose positional deviation and attitude angle are to be measured exists. The object A is substantially rectangular, and in FIG. 1, the object A is a rectangular object at the reference position.
Let the corners of A ₀ be P _{1 ,} P ₂ , P ₃ , and P ₄ , and the corners of object A in any posture be P ₁ ′, P ₂ ′, P ₃ ′, and P ₄ ′, and the above measurement Each side of object A is defined by straight lines l, m, s
and t. The measurement area for measuring the positional deviation and attitude angle of the object A includes a rectangular object A ₀ at the reference position.
An ITV camera or other appropriate image sensor V ₁ is arranged to place the intersection P ₂ of two orthogonal sides (straight lines l and m) approximately in the center of the field of view. This image sensor detects other corners _{P 1 ,} P ₃ or
It can also be placed on P ₄ . When the object A is transported by a belt conveyor or the like and reaches the measurement area, it usually has an arbitrary amount of positional deviation and an arbitrary amount of attitude angle inclination with respect to the reference position, as shown in FIG. There is. That is, the center of the object A is shifted by Δx and Δy in the x-axis and y-axis directions, respectively, and is tilted by an angle θ with respect to the reference posture of the object. Therefore, in order to measure the positional deviation and attitude angle, the image sensor _V1 detects the image of the corner of the figure formed by the intersection of the two sides of the object A, that is, the two straight lines l and m, and If the equations of the two straight lines are found, the positional deviation and attitude angle of the object can be measured by simple calculations. The present invention is for processing images from such an image sensor, etc., to obtain equations of two straight lines, and FIG. 2 shows the processing process in the form of a flowchart. In other words, when processing the above image, the image sensor
The image of object A taken into the image processing device by V ₁ is shown in the shape of the corner of a figure formed by the intersection of two straight lines, and this image information is first
The entire image is binarized based on the difference in brightness between the real part of the figure and the other parts. Next, from the binarized image, a spatial filter extracts a direction code indicating on which side of the n×n minute matrix the real part of the figure is located. FIG. 2 illustrates the relationship between a 2×2 mask pattern and the direction codes 0 to 9 attached to it. In this example, if the 2×2 area is all 0, code 0 means If it is 1, it is code 1, if only the left side is 1, it is code 2, if only the right side is 1, it is code 3, and hereinafter, the real part of the figure, that is, the binarized information, is shown on either side of the 2 × 2 matrix. Different codes are assigned depending on whether 1 is present. Although the case where the mask pattern is a 2×2 minute matrix has been described here, the mask pattern is not limited to this, and may be a 3×3 matrix, for example. After extracting such direction codes, as explained below, the total number of direction codes is added up for each combination of direction codes corresponding to the postures of the two straight line portions of the figure corner. After extracting the combined component of the direction code that shows the maximum value and determining whether or not two straight lines exist in the image, if there are two straight lines, among the combined components of the direction code, The components of the two straight lines are separated based on the difference in their corresponding direction codes. That is, as shown in Figure 1, the image sensor
When V ₁ is placed at the corner P ₂ of the object A ₀ , the direction code extracted from the binarized image by the spatial filter is listed in the P ₂ column of Table 1 according to the attitude angle θ of the figure. It becomes like that. Table 1 shows the corners P ₁ , P ₂ , P ₃ of the rectangular object A ₀
Or, for the case where the image sensor is placed at P ₄ , the attitude angle θ of the figure and the combination (α, β) of the direction code of two intersecting straight lines that has the maximum component sum within a certain angular range of the attitude angle θ. , α means a horizontal straight line component when θ=0, and β means a vertical straight line component when θ=0.

[Table] As can be seen from the description in column P ₂ of Table 1, when the attitude angle θ of the object in Fig. 1 is within the range of -90 to -45°, especially -90° The direction code combination (2, 5) is dominant on the side, and the direction code combination (8, 5) is dominant around -45°.
7) becomes dominant, so in the end, the above −90 to −
In the angular range of 45°, the direction code combination (2
+8, 5+7), and essentially there should be no other direction codes other than 0 and 1. Note that direction codes 2 and 8 are components on the straight line l, and direction codes 5 and 7 are components on the straight line l.
is the component on the straight line m. Furthermore, in the range of attitude angle θ from -45 to 0°, the combination of direction codes (8, 7) centered around -45° is dominant, and the combination of direction codes (8, 7) centered around 0° is dominant. Since the combination (4, 2) is predominant, essentially only (4+8, 2+7) exists in the above angle range of -45 to 0 degrees. Therefore, the number of components of each direction code in the entire screen is calculated individually, and then the direction codes 2, 8, which should be within the angular range of -90 to -45 degrees,
5,7 The respective directions for direction codes 4, 8, 2, 7, which should be in the angular range -45 to 0 degrees, and direction codes 4, 6, 2, 8, etc., which should be in the angular range 0 to 45 degrees. Calculate the total number of codes, and find the largest one among those totals. By finding the combination of direction codes that indicate this maximum value, the approximate attitude angle can be determined, and at the same time, the components of the direction codes of the two straight lines can also be determined. For example, if the total value of direction codes 4, 8, 2, and 7 is maximum, the attitude angle is within the angle range of -45 to 0 degrees, and the direction codes of the components of straight line l are 4 and 8, and the straight line It can be seen that the direction codes of the components of m are 2 and 7. In this way, find the combination of direction codes with the maximum total number, and include the straight lines l, m.
If it is detected that the components of The point sequence of can be separated and extracted. Then, the equations of the two straight lines can be calculated based on the separated directional code components. That is, as described above, the coordinates of each point constituting the point sequence are determined for each straight line, and the slope and intercept are determined by substituting these coordinate values into the equations of the straight lines l and m. For example, let the equation of straight line l be y = A ₁ x + B ₁ , and each point (x _1i , y _1i ) that makes up the sequence of points on straight line l
Substituting into it, we get the average slope A ₁ and intercept B ₁
If you ask for given by. The equation of the straight line m can be found in the same way. In this manner, according to the above method, two straight lines in an image can be separated by performing the above-mentioned simple processing with the arithmetic processing device connected to the image sensor _V1 or the like. [Effects of the Invention] According to the method of the present invention, an image sensor or the like is placed at the corner of the object, and based on the output image, a linear equation for two straight lines in the image can be easily and accurately calculated. You can ask for it. In addition, since measurement can be performed easily in this way, it is not only easy to use a device for processing images as a NerdWare device, but also real-time processing can be performed.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram showing an application example of the method for separating two straight lines according to the present invention, Fig. 2 is a flowchart showing the process of processing by the method of the present invention, and Fig. 3 is a 2
FIG. 2 is an explanatory diagram of the relationship between the mask pattern and the direction code. V ₁ ...Image sensor, A...Object.

Claims (1)

[Claims] 1 After the image of the corner of the figure formed by the intersection of two straight lines obtained by an image sensor etc. is completely binarized, it is converted to n×n by a spatial filter.
Extract the direction codes that indicate which side of the micro matrix the real part of the figure is on, and calculate the total number of direction codes for each combination of direction codes that correspond to the attitude angle of the straight line part at the corner of the figure. The combined components of the direction codes that show the maximum value are extracted, the direction codes corresponding to the two straight lines are separated from the combined components, and the above is calculated based on the data corresponding to each separated direction code. A method for separating two straight lines in an image, the method comprising calculating a linear equation for the two straight lines.