JP2558096B2 - Automatic measurement method of posture and size of object with straight line part - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention relates to an object, in particular, four substantially straight rectangular parts for assembly of parts in a production line of automobiles and home appliances. The present invention relates to a method for automatically measuring a positional deviation, an attitude angle, and a dimension of an object having

[Prior Art] Various types of pattern recognition technology have been conventionally developed in order to automate work on a production line. Such pattern recognition technology attempts to determine the shape, posture, etc. of the object itself by some means, and the device used for pattern recognition is usually relatively complicated and expensive. Is.

However, the pattern recognition technique as described above is not always necessary in the assembly work in various factories, for example, in the case of automating the assembly of various parts to the body of the automobile or the body of the home electric appliance. That is, the body of the automobile, the body of the home electric appliance, the IC substrate, etc., which is the object of recognition or the like, usually has a substantially rectangular shape, and even if the recognition object itself is not rectangular, it has a rectangular shape. Like to move it on a dolly,
In many cases, a rectangular portion exists on the object. Moreover, in the case where the object has a rectangular shape in this way, even if a large number of kinds of objects to be recognized are mixed, the shape can be identified by measuring the vertical and horizontal dimensions. .

Therefore, it is sufficient for the recognition of the object to accurately measure the displacement of the substantially rectangular portion with respect to the reference position, the inclination of the posture angle with respect to the reference posture, and the vertical and horizontal dimensions. It can be significantly simplified and provided at a low cost.

[Problems to be Solved by the Invention] When recognizing the posture and dimensions of the rectangular object described above, the present inventors use a two-dimensional image of the object obtained by an ITV camera or other image sensor as much as possible. The present invention has been accomplished by studying a method of performing accurate measurement with a simple arithmetic process and using the smallest possible number of image sensors.

That is, an object of the present invention is to pay attention to the fact that the recognition object is usually rectangular, and in the assembly work in the above-mentioned factory, etc., without using general pattern recognition technology, It is possible to accurately measure the positional deviation, the attitude angle, and the dimension by a simple process of the image of the image sensor.

[Means for Solving the Problems] In order to solve the above problems, in the automatic measuring method of the present invention, the average at the reference position is set in the measurement area for measuring the posture and dimensions of the substantially rectangular object. Arrangement of an image sensor that fits within the visual field the central portion in the longitudinal direction of each side of the object of a rectangular shape of the objective size, each of these image sensors, each side of the object of the average size, It is assumed that the width of the field of view is kept approximately at the center of each field of view, and the image of the measurement object obtained by each image sensor is binarized by the difference in brightness, and each side of the circumference of the measurement object is binarized. The point sequence on the straight line is extracted, and from the coordinates of each point constituting the point sequence in the Cartesian coordinate system having the origin at the center of the object of the average size at the reference position, the Equation of each side And calculate
The technical means of measuring the positional displacement of the center of the measurement target with respect to the center of the coordinate system, and the posture angle and the dimension of each side of the measurement target by calculation based on these equations is adopted.

[Examples] The method of the present invention will be described in more detail below with reference to the drawings.

As shown in FIG. 1, it is assumed that an object A ₀ having an average size is present at a reference position in a space where the object A whose posture and size are to be measured is present, and an origin 0 is located at the center of the object xy.
Set the Cartesian coordinate system. As the measurement object A, a plurality of kinds of objects having different vertical and horizontal dimensions are mixed, and all of them are substantially rectangular. The rectangular object may be, for example, a two-dimensional view of the body of an automobile, or the body of an electric home appliance, an IC substrate, or the like. Furthermore, if the target object is not rectangular,
The measurement target A may be a rectangular portion attached thereto or a rectangular table on which the object is placed. In addition, the fact that a plurality of types of measurement objects A having different vertical and horizontal dimensions are mixed means that, for example, the width W and the length of a vehicle body of an automobile sequentially sent to the measurement area for assembly of parts and the like. This means that a plurality of vehicle bodies having different dimensions for H are mixed.

In FIG. ₁ , the corners of the rectangular object A ₀ having an average size at the reference position are P ₁ , P ₂ , P ₃ and P ₄ , and the corners of the object A having an arbitrary posture are ▲ P ^′ _1. ▼, ▲ P ^' ₂ ▼, ▲ P ^' ₃ ▼
And ▲ P ^′ ₄ ▼, and each side of the measurement object A is indicated by straight lines l, m, s and t.

In the measurement area for measuring the displacement, posture angle and size of the object A, a rectangular object having an average size at the reference position.
ITV for each side of A ₀ (straight lines l, m, s, t) that keeps the central part of each side in the longitudinal direction within the field of view
A camera and other appropriate image sensors V ₁ , V ₂ , V ₃ , V ₄ are arranged.

Each of these image sensors V ₁ , V ₂ , V ₃ , V ₄ is arranged so that each side of the object A ₀ having the above-mentioned average size is approximately centered in each field of view and is held over the entire width of the field of view. To do. By arranging each image sensor in this way, even if there is a displacement of the object or a change in the posture angle, the straight line parts on each side can be maximally placed within the field of view and accurate measurement can be performed. .

When the object A is transported by a belt conveyor or the like and reaches the measurement range, it usually has an arbitrary amount of positional deviation with respect to a reference position and an arbitrary amount of posture angle inclination, as shown in FIG. . That is, the center of the object A is Δx and Δy in the x-axis and y-axis directions with respect to the origin 0 of the coordinate system.
It has a positional deviation of 10 mm, and is inclined by an angle θ with respect to the reference posture of the object.

Therefore, in order to measure the displacement, the posture angle and the dimension, the image sensors V ₁ , V ₂ , V ₃ and V _{4 are used} to measure the respective sides of the object A, that is, the straight lines l, m, s and t. By detecting the position as an image, calculating the equations of these straight lines, and performing the calculation as described below, the origin 0 of the coordinate system can be obtained.
The positional deviations Δx, Δy of the center of the object A in the x-axis and y-axis directions, the attitude angle θ with respect to the reference attitude, and the width W and the length H of the object can be obtained.

First, the image sensor V _1, V _2, V _3, V ₄ of line l around the sides of the object A based on the output, m, s, equations t has an image sensor V _1, V _2, V ₃ , V ₄ is binarized by the difference in brightness between the rectangular object A and other portions, and appropriate processing such as differentiation is performed to extract the point sequence on the straight line. It can be obtained by obtaining the coordinates of each point constituting the point sequence and substituting the coordinate values into the equation of the straight line l, m, s, t to obtain the slope and intercept.

That is, the equation of the straight line l is set to y = A ₁ x + B _1, and each point (x _1i , y _1i ) forming the point sequence on the straight line 1 is substituted into it, and the average gradient A ₁ and the intercept B ₁ And ask
They are, Given by.

Similarly, the equation of the straight line m is set as y = A ₂ x + B _2, and each point (x _2i , y _2i ) forming the point sequence on the straight line m is substituted into it, and the average gradient A ₂ and intercept If we ask for B ₂ ,
They are, Given by.

Further, the coordinates of the corner portion ▲ P ^′ ₂ ▼ of the object A whose position is deviated from the corner portion P ₂ of the object A ₀ are Represented by

Therefore, the straight line l has an inclination of the angle θ, and ▲ P ^′ ₂ ▼
Since the equation of the straight line l is obtained from the point passing, Therefore, the slope A ₁ and the intercept B ₁ are A ₁ = tan θ (2) Is represented.

Similarly, the equation of the straight line m is And the slope A ₂ and intercept B ₂ are A ₂ = −tan θ Is represented.

Therefore, from the above equation (2), θ can be obtained by θ = tan ⁻¹ (A ₁ ) ... (5), and based on the intercept of the equation expressing the straight lines l and m, that is, the above (3) From equation (4), As a result, Δx and Δy can be obtained.

However, since these equations (6) and (7) include unknown W and H, it is necessary to eliminate them.

Therefore, the outputs of the image sensors V ₃ and V ₄ are set to W and H
It will be used to measure. Also in this case, similarly to the case described above, the coordinate values (x _3i , y _3i ) (x _4i , y _4i ) of the points forming the point sequence on the straight lines are added to the equations of the straight lines s and t. By substituting into it the average gradients A ₃ , A ₄ and the intercepts B ₃ , B ₄ and from the relationship with the coordinates of the corners ▲ P ^′ ₄ of the object A, for the straight line s, Get.

Similarly, for the straight line t, Get.

Then, by (8) × tan θ− (9), (8) + (9) tan θ Is obtained, and Δx is eliminated from the equations (6) and (10), Eliminating Δy from equations (7) and (11), Get.

Where C = cos θ, D = sin θ, Then, the above formulas (12) and (13) are expressed as C · W−D · H + E = 0 D · W + C · H + F = 0. From these, it is possible to obtain H and W by the following formula. it can.

The values of these H and W are described in (6) and (7) above.
The values of Δx and Δy can also be obtained by substituting into the equation.

In the above, Δx and Δy are calculated based on the intercepts of the equations representing the straight lines l, m, s, and t.
The calculation can also be performed from the intersection coordinates of l, m or the straight lines s, t. For example, if the coordinates of the corner ▲ P ^′ ₂ ▼ of the object which is the intersection of the straight lines l, m are (x ₁ , y ₁ ) and the coordinates of the corner ▲ P ^′ ₄ ▼ are (x ₂ , y ₂ ). , Their coordinate values are And based on these, Δx, Δy, W
And H can also be calculated.

As described above, by performing the above-described calculation with the arithmetic device connected to the image sensors V ₁ , V ₂ , V ₃ , and V ₄ , the positional deviations Δx and Δy of the object A in the x-axis and y-axis directions, The posture angle θ with respect to the reference posture and the dimensions H and W of the object can be easily obtained.

[Effect of the Invention] According to the measuring method of the present invention as described above, the displacement of the object, the posture angle, and the dimension of the object can be achieved by a simple means of performing the calculation of the above formulas based on the outputs of the four image sensors. Can be accurately measured. In particular, since the image sensor is used only to obtain the equation of the straight line of each side on the object, the image sensor is used to observe the entire object for pattern recognition or According to the similar technique, not only the calculation processing becomes extremely simple, but also extremely accurate measurement can be performed, as compared with the case where the position shift, the posture angle and the dimension of the object are measured.

Moreover, since each image sensor is arranged so as to correspond to substantially the center in the lengthwise direction on each side of the object, and each side is arranged so as to fit over the entire width of the field of view in the substantially center of the field of view, Even if there is a deviation or a change in posture angle, the straight line parts on each side are maximized within the field of view, and the edges of each side are less likely to be in the field of view. It becomes possible to measure.

In addition, since the measurement can be performed easily in this way, not only is it easy to implement a device for processing the output from the image sensor as hardware, but also real-time processing is possible, and therefore, in a production line of a factory or the like. Very suitable for use.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a measuring method according to the present invention. V ₁ , V ₂ , V ₃ , V ₄ …… Image sensor, A …… Object.

 ─────────────────────────────────────────────────── ─── Continuation of the front page Jury Chairman Shun Terao Judge Judge Katsumi Enari Judge Yoshiyuki Shimonaka (56) References JP-A-53-39156 (JP, A) JP-A-53-123957 (JP, A) JP-A-53-145526 (JP, A) JP-A-58-58405 (JP, A)

Claims (1)

(57) [Claims] 1. A measurement area for measuring the posture and dimensions of a substantially rectangular object includes a central portion in the longitudinal direction of each side of a rectangular object having an average dimension at a reference position. An image sensor that fits within the field of view is provided, and each of these image sensors is designed to hold an object of the above average size on each side at approximately the center of each field of view over the entire width of that field of view. The image of the measurement object obtained by the sensor is binarized by the difference in brightness, and the point sequence on the straight line of each side of the circumference of the measurement object is extracted to obtain the average size of the average position at the reference position. From the coordinates of the points that make up the point sequence in the Cartesian coordinate system that has the origin at the center of the object, calculate the equations for each of the above-mentioned peripheral sides on the object, and then use the equations to calculate the coordinate system. Total to the center of Positional deviation of the center of the object, as well as to measure the attitude angle and dimensions of each side of the measurement object, orientation and automatic measurement method of the dimensions of the object having the straight portion, characterized in that.