JP2966667B2 - Shape estimation method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an object having a bent portion, which is based on a point sequence in which the positions of a large number of measurement points arranged on a line of intersection with a cutting plane crossing the bent portion are represented by two-dimensional coordinates.
The present invention relates to a shape estimation method for approximating the shape of an object on the intersection with a polygonal line.

[0002]

2. Description of the Related Art Generally, in order to measure the shape of a bent portion of an object having a bent portion, it is necessary to specify a bent position. For example, when welding a pair of plate materials to form a square joint, it is necessary to measure a step or a gap formed between the two plate materials. It is necessary to specify a bending position for the formed bent portion.

[0003] In order to satisfy such a demand, the positions of measurement points are represented by two-dimensional coordinates with respect to a large number of measurement points on a line intersecting with a cutting plane traversing a bent portion of an object, and a point sequence on a coordinate plane is represented. There has been proposed a method of extracting a point (referred to as a bending point) of a bending position of an object from the above, thereby approximating a shape of the bending portion of the object on the intersection line with a straight line connecting the bending points. As a method for obtaining position information of a measurement point on a line of intersection with a cutting plane, a light cutting method is generally known.

As a method of extracting a bending point from a point sequence on a two-dimensional coordinate plane and approximating the shape of the point sequence by a polygonal line connecting the bending points, a dividing / synthesizing method (T. Pavlidis and S. et al.
L. Horowitz, “Segmentation of Plane Curves,” IEE
E Trans. Comput., Vol. C-23, pp. 860-870, August 1
974) and tracking methods (J. Sklansky and V. Gonzalez, “FastP
olygonal Approximation of Digitized Curves, ”Patt
ern Recognition, vol. 12, pp. 327-331, 1980).

In the division / synthesis method, as shown in FIG.
Each point P _j (j included in the point sequence S to be a polygonal line approximation
.., N) are arranged in order from the left, and a line segment <P ₀ P _n > connecting the start point P ₀ and the end point P _{n as} shown by a solid line in FIG. And set the line segment <P ₀ P
_Then, a Euclidean distance E _j between the _n > and each point P _j is calculated, and a point P _jEMAX at which the distance E _j is maximum is obtained. Here, the line segment <P ₀ P _n> a point when the distance E _JEMAX the P _JEMAX exceeds a predetermined threshold value E _th, 2-part point sequence S ₁ point sequence S before and after the point P _JEMAX, S Divide into _two .

The partial point sequences S ₁ , S ₂ thus obtained
, The same operation as that of the original point sequence S is performed. That is, as shown in FIG. 18B, each partial point sequence S ₁ ,
_Line segment <P ₀ P _jEMAX >, <P _jEMAX P _n > for S ₂
Is set, and each point P _j included in each of the partial point sequences S ₁ and S ₂ is set.
Each line segment _{<P 0} P jEMAX>, after obtaining the Euclidean distance between the _<P jEMAX P _n>, each segment _{<P 0} P jEMAX> and,
<P _jEMAX P _n > and the point P where the Euclidean distance becomes maximum
_jEMAX1, P _jEMAX2 distance between the E _jEMAX1, E _jEMAX2 the threshold E
_It is determined whether or not it exceeds _th, and it is determined whether or not the partial point sequences S ₁ and S ₂ are further divided.

As described above, the division is repeated until the maximum value of the distance between the line segment connecting the start point and the end point and the point included in the partial point sequence does not exceed the threshold value E _th for each partial point sequence. The point sequence S can be approximated by a polygonal line, which is a set of line segments connecting the start point and the end point of each partial point sequence. That is, in the example of FIG. 18, part point sequence S _11, S as only partial point sequence S ₁ in FIG. (B) is the figure (c)
An example of division into ₁₂ is shown, and the state in FIG. 10C shows a state in which the maximum values E _jEMAX11 and E _jEMAX12 of the Euclidean distance are smaller than E _th and the division is completed. Therefore, the point sequence S is composed of three line segments <P ₀ P _jEMAX1 >, <P
_jEMAX1 P _jEMAX>, are polygonal approximated by _<P jEMAX P _n>.

On the other hand, in the tracking method, as shown in FIG.
Each point P _j (j included in the point sequence S to be broken line approximation
= 0, 1,..., N) are arranged in order from the left, and the start point P ₀ is set as one end point P _j1 and the other points P _j included in the point sequence S as shown in FIG. _{Is set} as the other end point P _j2 , and a line segment <P _j1 P _j2 > connecting both end points P _j1 and P _j2 is set. Here, two tangents λ _{1, j2-1} , λ ₂ passing through the end point P _j1 among the tangents of the circle C _j2-1 having the radius ε centered on the point P _{j2-1 immediately} before the end point P _{j2. , j2-1} , a line segment <P _j1 P _j2 > is sandwiched between the point P _j2-1 and the line segment <P _j1 P _j2 >.
It is as follows. Therefore, one end point P _j1 is fixed, and the other end point P _j2 is changed in the arrangement order of the point sequence S, and the point P
It is determined whether the distance between _j2-1 and the line segment <P _j1 P _j2 > is equal to or smaller than ε. Thus, the point P _j2-1 and the line segment <P _j1
Longest line segment <P _j1 P _j2 > whose distance from P _j2 > is equal to or less than ε
Choose out. That is, in the example of FIG. 19, as shown in FIG. 19B, the distance between the point P _j2-1 and the line segment <P _j1 P _j2 > is ε from J1 = 0, J2 = 2 to J2 = 8. Although the following condition is satisfied, the condition is no longer satisfied when J2 = 9, so that the line segment <P ₀ P ₈ > is set as the longest line segment.

Next, the final end point P satisfying the above conditions
_j2 (P _{8 in the} above example) is set as an end point P _j1, and the longest line segment <P _{j1 in} which the distance between the point P _j2-1 and the line segment <P _j1 P _j2 > is equal to or less than ε in the same procedure. P _j2 >. That is, FIG.
In the example of FIG. 9, J1 = 8 and J2 = 1 as shown in FIG.
The above process is repeated starting from 0. In this manner, the process of selecting the line segment <P _j1 P _j2 > until the end point P _j2 reaches the end point P _n of the point sequence S is performed.
As shown in (d), the point sequence S can be approximated by a polygonal line connecting the line segments <P _j1 P _j2 >.

[0010]

The object of the invention is to be Solved by the way, amount corresponding to the above-mentioned
When approximating the point sequence S by a polygonal line by the dividing / synthesizing method, if the optimum value is selected as the threshold value E _th , there is a possibility that almost ideal approximation can be performed. However, the point including (n + 1) points P _j In the case of extracting Nb inflection points from the column S, the calculation for obtaining the distance between the line segment connecting the start point and the end point of each partial point sequence and the point included in each partial point sequence is performed at least (n + 1)
× Nb times. Therefore, there is a problem that the time required for the calculation becomes very long.

On the other hand, in the tracking method, assuming that the point sequence S includes (n + 1) points, the equation of the line segment <P _j1 P _j2 > and two tangents λ _{1 of the} circle C _j2-1 are used _{. Since the} calculation of the formulas _j2-1 , λ2 _{, and j2-1} only needs to be performed (n-1) times, the advantage is that the amount of calculation is small and the time required for calculation is short as compared with the division / synthesis method. Have. However, when the distance between the point P _j2-1 and the line segment <P _j1 P _j2 > exceeds ε, the point P _j2-1
Is the inflection point, so as shown in FIG.
Next point P ₈ beyond P ₇ points to be selected as the bending point may become bent point, there is a problem of low reliability of the bending points obtained.

SUMMARY OF THE INVENTION An object of the present invention is to solve the above-mentioned problem, and to provide a shape estimating method capable of appropriately determining the position of a bending point in a relatively short time when approximating a point sequence by a polygonal line. It is assumed that.

[0013]

According to the first aspect of the present invention, in order to achieve the above object, the position of a large number of measurement points on an object having a bent portion, which is arranged on a line of intersection with a cutting plane traversing the bent portion. In the shape estimation method of approximating the shape of the object on the intersection with a polygonal line based on a point sequence in a predetermined order represented by two-dimensional coordinates, a predetermined number of points are separated from each other in the point sequence. The cosine of the angle between the two straight lines passing through the two points is determined as the degree of bending, and the point where the degree of bending is maximized in the vicinity range set for each point is defined as the bending point, and is formed by a line segment connecting the bending points in order. Find a broken line.

According to the second aspect of the present invention, for only the points included in the sequence of points whose degree of curvature exceeds a predetermined threshold value, the point where the degree of curvature is maximum in the vicinity of the point is determined as the inflection point. . According to the third aspect of the present invention, when the length of the line segment connecting the adjacent bending points is shorter than a predetermined lower limit after obtaining the line segment connecting the bending points, of the bending points at both ends of the line segment, Are excluded as unnecessary bending points.

According to the fourth aspect of the present invention, after the line connecting the inflection points is obtained, if the angle between the two lines passing through the inflection points is larger than a predetermined upper limit, the inflection points are unnecessary. Is excluded as an inflection point. In the invention of claim 5, after determining the inflection point, a distance between a line segment connecting two adjacent bending points on both front and rear sides of the target inflection point and a target inflection point is obtained,
When this distance is smaller than a predetermined threshold, the target bending point is excluded as an unnecessary bending point.

[0016]

According to the present invention, the degree of bending of a point sequence is evaluated by determining, as a degree of bending, the cosine of the angle between two straight lines passing two points separated by a predetermined number of points before and after each point included in the point sequence. And, when the degree of bending in the vicinity of each point is maximized, it is determined to be a bending point corresponding to the bending portion,
Compared with the tracking method described in the conventional example, the bending point can be accurately obtained, and the appropriate bending point accurately reflecting the shape of the object can be obtained.

In the method for setting a threshold value for the degree of bending as described in claim 2, it is possible to remove a minute fluctuation caused by a measurement error or the like, so that erroneous detection due to the influence of the measurement error can be suppressed. it can. Further, as described in claims 3 to 5, if an unnecessary bending point is removed after obtaining the bending point, a more appropriate bending point can be obtained. In addition, if the original point sequence has a large number of straight lines, the number of inflection points to be evaluated as to whether or not it is unnecessary can be greatly reduced by removing the error component using the threshold as described in claim 2. Even if processing for removing unnecessary bending points is added as in claims 3 to 5, the amount of calculation is reduced as compared with the conventional division / synthesis method.

[0018]

【Example】

(Embodiment 1) In this embodiment, as shown in FIG. 16, the light source 2 irradiates the object 1 with slit light traversing the bent portion of the object 1, and the irradiation direction of the slit light from the light source 2 is as follows. By imaging the projection pattern of the slit light from an image pickup device 3 such as a CCD camera from different directions, and performing the following processing on the image captured by the image pickup device 3, a point sequence corresponding to the shape of the object 1 is obtained. Extract. In other words, the position of the measurement point is represented by two-dimensional coordinates with respect to each measurement point on the line of intersection with the cut surface traversing the bent portion of the object 1 by the so-called light section method.

The grayscale image Q ₁ (FIG. 17A) captured by the imaging device 3 is input to the binarization processing section 11. The binarization processing unit 11, the density of each pixel is compared with the appropriate threshold to obtain a binary image Q ₂ as shown in FIG. 17 (b) by the density of each pixel is binarized. Next, the noise component of the binary image Q ₂ is removed by the noise removing unit 12. Here, the binary image Q
_Since the line figure I ₂ obtained in ₂ generally has a width of a plurality of pixels, the skeleton extraction unit 13 extracts a skeleton as the center line of the line figure I ₂ in the binary image Q ₂ , as shown in FIG. 17 (c), to obtain a line image Q ₃ consisting of data line figure I ₃ having a width of 1 pixel. Line figure I
₃ is a column of pixels having a width of one pixel,
Each pixel of ₃ is a dot sequence reflecting the shape of the bent portion of the object 1. Therefore, the position data of the point sequence by aligning the position coordinates of each pixel of a line figure I ₃ from the left on the horizontal direction of the screen in order. Here, although all pixels in line figure I ₃ is set to point sequence, if may be lowered resolution or obtain the position coordinates of one pixel every proceeds plurality of pixels in the horizontal direction of the screen, for each of the plurality of pixels The representative value may be used as the position coordinates. Thus, based on the line figure I ₃ , FIG.
It is possible to obtain a point sequence S as shown in FIG. The position data of the point sequence S is stored in the storage unit 14. The points included in the point sequence S obtained in this manner have an order from the left in the horizontal direction of the screen.

Next, the polygonal line processing unit 20 approximates the arrangement of the point sequence S with a polygonal line. The polygonal line processing unit 20 evaluates the point sequence S in the following procedure. Here, it is assumed that the point sequence S is composed of points P _j (j = 0, 1,..., N) arranged in order from the left as shown in FIG. First, for each point P _j , the degree of curvature Cv _j is obtained in the degree of curvature calculation unit 21.
The degree of bending Cv _j is defined as follows. First, FIG.
As shown in an enlarged view of a portion surrounded by a broken line) of, d the d items (Figure 3 before and after the arrangement order in the vicinity of each point P _j
= 5) The points P _jd and P _{j + d} at the shifted positions are obtained, and the degree of bending C
Angle θ _j formed by two line segments <P _j P _jd > and <P _j P _{j + d} > connecting point P _j for _obtaining v _j and points P _jd and P _{j + d}
Ask for. Since the angle θ _j takes a value in the range of [0, π], the cosine cos θ _j of the angle θ _j monotonically decreases in the range of [−1, 0]. Therefore, if the degree of flexion Cv _j is defined as cos θ _j , a corresponding relationship is obtained in which the greater the value of the degree of flexure Cv _j , the greater the degree of flexion near the point P _j .

When calculating the degree of flexion Cv _j , <P _j
P _jd > and <P _j P _{j + d} > are each a vector → D
Assuming that p _j , → Dn _j (hereinafter, the symbol preceded by → indicates a vector. However, in the figure, → is added to the symbol to represent the vector), the vector → Dp _j , → Dn _j is used. That is, the point P _j
Assuming that the position vector from the origin of (x _j, y _j ) is → Pv _j , the position vector → Pv _j is calculated as follows using the horizontal unit vector of the screen → u and the vertical unit vector → v. Can be expressed as

→ Pv _j = x _j (→ u) + y _j (→ v) Therefore, the vectors → Dp _j , → Dn _j are as follows. → Dp _j = → P _jd − → Pv _j = (x _jd −x _j ) (→ u) + (y _jd −y _j ) (→ v) → Dn _j = → P _{j + d} − → Pv _j = ( x _{j + d} −x _j ) (→ u) + (y _{j + d} −y _j ) (→ v) Further, by using the absolute value and inner product of each vector, both vectors → Dp _j , → Dn _j cosine cos θ _j of the angle θ _j
Is as follows:

Cos θ _j = (→ Dp _j. → Dn _j ) / (│ → Dp _j │ ・ │ → _{D n j} │ After all, based on the coordinates of each point P _j , P _jd , P _{j + d} , The cosine cos θ _j can be obtained, and the cosine cos θ _j thus obtained can be used as the degree of curvature Cv _j . When the degree of curvature Cv _j is obtained for each point P _j ,
The bending point extraction unit 23 evaluates the degree of bending Cv _j of each point P _j and determines whether the point P _j is a bending point. In bending point extraction unit 23, as shown in FIG. 4, s is the s (Fig before and after the point P _j
= 2), the neighborhood range [P _js , P
_{j + s} ]. The neighborhood range _{[P js, P j + s} ] P js points included in, ..., P _j, ..., tortuosity Cv _js in _{P j + s, ..., Cv} j, ..., is Cv _{j + s} If there is a maximum point, the point is _defined as a bending point Bpk (k = 1, 2,...)
And Here, the starting point P ₀ and the ending point P _n of the point sequence S are regarded as bending points. In addition, the neighborhood range [P _js , P
The value of s for setting the magnitude of [ _{j + s} ] is appropriately selected based on the interval between the bending points in the point sequence S.

By the way, the range [P _js , P _j
even if the _{j + s]} with the degree of curvature is maximum, because if the bending degree is small is the hardly bent, such point it is desirable to have excluded from the bending point Bp _k. Therefore, the threshold value processing unit 22 sets a threshold value Tc relating to the degree of bending, and only when the degree of bending exceeds the threshold value Tc, the bending point extracting unit 23 determines whether or not the bending point is maximized. Here, an expected value D _max for the maximum length of the vector → Dp _j , → Dn _j set when the degree of curvature Cv _j is obtained, and an allowable value of the distance from the polygonal line approximating the point sequence S to the point P _j If ε (shown in the conventional example) is used, since sin θ _j <ε / D _max , the following value can be adopted as the value of the threshold Tc.

Tc = {1− (ε / D _max )} ^1/2 By evaluating the degree of flexion Cv _j using such a threshold value Tc, a portion that can be regarded as a straight line and a portion that can be regarded as a bent portion are obtained. It can be separated and the frequency of occurrence of unnecessary bending points can be reduced. In this way, the bending point Bp _k obtained at the bent point extracting unit 23 segments Se
_The point sequence S can be approximated by a broken line if they are connected in order by _m (m = 1, 2,..., k−1).

After all, in this embodiment, as shown in FIG.
The bending degree Cv _j is obtained for each point P _j of the point sequence S (S
1), it is then tortuosity Cv _j point is the maximum value P _j taken as the bending point Bp _k (S2). In FIG. 1, the process in step S3 is a process corresponding to the second to fourth embodiments described later. (Example 2) In the embodiment described above, the vicinity range [P _js to set in order to determine the point at which the degree of curvature Cv _j is maximum, P
_{j + s]} or set large, or when you set large the threshold value Tc for evaluating the degree of curvature Cv _j, because there may not be extracted inflection point Bp _k to be extracted, the neighborhood area [P _js, P _{j + s} ] and the threshold value Tc need to be set small. However, the neighborhood range [P _js ,
Setting small P j _{+ s]} and the threshold value Tc, so that the problem of unnecessarily many bending points Bp _k are extracted results. Therefore, it is desirable performs processing to further remove unwanted bending point after obtaining the inflection point Bp _k.

[0027] In the present embodiment, after obtaining the bending point Bp _k at the bent point extraction unit 23, the unnecessary point removing unit 24,
Unnecessary bending points are removed based on the length of the line segment. First,
As shown in FIG. 6 (a) (hereinafter, omitted from describing only relationship bending point Bp _k, from FIG other points except the bending point Bp _k among the points P _j contained in the point sequence S) , to seek length Sl _m line segments Se _m determined as in example 1, to set the lower limit Sl _th this length Sl _m.
Here, as shown in FIG. 5, if there is a line segment Se _m as the _{Sl m <Sl th (S30)} , the bending point Bp _m at both ends of the line segment Se _m, to remove one of the Bp _{m + 1} ( S31). Lower limit Sl _th is permissible value of the distance from the polygonal line approximating the point sequence S to the point P _j epsilon (described above) is set to about.

[0028] Both of the bending point Bp _m, or to remove Bp m + ₁ is determined as follows. That is, FIG.
(B) as in (Fig. (An enlarged view of a portion surrounded by a broken line in a)), the bending point Bp _m, segment Se _m-1 of the two sandwiching Bp _m + _1, Se m , Se _m , and Se _{m + 1} , the angles ψ _m-1 and ψ _m (m = 4 in the figure) sandwiching them are obtained, and the bending points Bp _m , where the angles ψ _m-1 and ψ _m are larger, are obtained. Remove Bpm _{+ 1} . For example, ψ _m-1 >
If [psi _m removes bending point Bp _m. In this way Bp bending point is shown in FIG. _{6 (a) k (k =} 0,1,
FIG. 7 shows the result of removing unnecessary bending points Bp ₂ from..., 9). Here, as a result of bending point Bp ₂ is removed, it becomes bent point Bp ₃ newly bending point Bp _2, below, so that the order of the bending point is increased repeatedly. Other points are the same as in the first embodiment. In this embodiment, the inflection points at one end of the line segment are removed, but the inflection points at both ends of the line segment may be removed.

(Embodiment 3) In this embodiment, as shown in FIG. 9 (FIG. 9B is an enlarged view of a portion surrounded by a dashed line in FIG. 9A), the bending point Bpm _{+ 1} is Line segment S
e _m, (in the figure is set to m = 3) angle [psi _m sandwiching the Se m + ₁ to set the upper limit value [psi _th for. As shown in FIG. 8, in the unnecessary point removing unit 24, the bending point B where ψ _m > ψ _th is satisfied.
If there is p _{m + 1} (S32), the bending point Bp _{m + 1} is removed (S33). Here, the upper limit ψ
_th is determined based on a point sequence S in such allowable value of the distance from the expected value of the length of a line obtained when the polygonal line approximation or each line segment to the point P _j epsilon. Therefore, the method of this embodiment is effective when the length of a line segment obtained when each point sequence S is approximated by a broken line can be predicted in advance. In this way, the bending point Bp _k (k = 0,
FIG. 10 shows the result of removing unnecessary bending points Bp ₄ from 1,..., 8). Other points are the same as in the first embodiment.

[0030] By employing the method of (Example 4) Example 2 and Example 3, after obtaining the bending point Bp _k, since it is possible to remove unwanted bending point, determine the bending point Bp _k treatment There is an advantage that can be simplified.
However, when removing the unnecessary bending point Bp _k, To prevent removal to the bending points required, the length of the lower limit value Sl
_th is relatively small it is desirable to set, and the upper limit value [psi _th angle is desirably set to a relatively large. As a result, unnecessary bending points may remain.

Therefore, in the present embodiment, the following method is employed as a method for removing unnecessary bending points more reliably. That is, since it can be assumed that the partial point sequence arranged between two adjacent bending points in the point sequence S is substantially aligned, the three adjacent bending points B
p _k-1, Bp _k, the line segment in _{Bp k + 1 <Bp k-} 1 B
that distance to p k _{+ 1>} is maximized becomes bending point Bp _k. Based on this finding, the unnecessary point removing unit 24, as shown in FIGS. 11 and 12 (FIG. 12B is an enlarged view of a portion surrounded by a dashed line in FIG. bending point adjacent the _{bp k bp k-1, bp} k + 1
The seek distance E _k between the line segment _{<Bp k-1 Bp k +} 1> connecting (seeking distance E ₆ the bent point Bp ₆ and segment <Bp ₅ Bp _7> in the figure), the distance when E _k exceeds the allowable value ε of the distance is to exclude bending point Bp _k as required point. That is, as shown in FIG.
The number of _k is initialized (S34), and the distance E
_k is _obtained (S35). When the distance E _k exceeds the allowable value ε of the distance (S36), excludes the bending point Bp _k (S37). Nb the maximum value of all the bending point Bp _{k (k}
) By performing the above processing (S
38, S39), it is possible to remove unwanted bending point Bp _k. Figure 12 bending point Bp _k shown in (a) this way (k = 0,1, ......, 7 ) shown in FIG. 13 the result of removing the unnecessary bending point Bp ₆ from. Other points are the same as in the first embodiment.

The above Examples 2 to 4 is an various embodiments of how to exclude unwanted points after obtaining the bending point Bp _k, the method of these examples alone or in combination employed can do. In the example shown in the figure, the methods of the second, third, and fourth embodiments are sequentially applied to the same data.
It can be seen that unnecessary points are sequentially removed as shown in FIG.

If the point sequence S is approximated by a polygonal line as described above, the shape of the object 1 can be measured as follows. Here, a pair of plate materials M ₁ , M ₂
An example of measuring the positional relationship and the like between the two plate materials M ₁ and M ₂ when forming a square joint by welding the two will be described. Here, as the data of the polygonal line approximation, a result obtained by sequentially applying the corrections of the second and third embodiments to the result obtained in the first embodiment is used. When the shape of the object 1 is superimposed on this result, the result is as shown in FIG.
Is obtained by superimposing the shape of the object 1 on the result of (1).
It is assumed that the plate thickness t of the plate materials M ₁ and M ₂ is known.

As shown in FIG. 14, the determined bending point Bp_k
Each plate material M constituting the object 1 from among₁, M_TwoCorner point F
₁, F_TwoIs selected (S11). next,
Left corner point F₁The data on the left side is the plate material M on the left side.₁of
Corresponds to top surface, right corner point F _TwoData to the right of
Side plate material M_TwoThis is considered to be equivalent to the upper surface of
About these data₁, M_TwoLine approximating the upper surface of
A minute is set (S12). Further, the right side plate material M_Twoupon
Surface extension line and left corner point F₁And the distance A
If the sheet thickness t is subtracted from the separation A, both sheet materials M₁, M_TwoBetween
This means that the gap g has been obtained (S13). Also, the left plate
Material M₁Extension of the upper surface of the right corner point F_TwoFind the distance B with
In other words, this distance B corresponds to the step of the square joint (S1
4), both plate materials M₁, M_TwoThe intersection C of the extension of
The position of the joint is determined (S15). As above
To obtain the dimensions required for welding square joints.
Because

[0035]

According to the present invention, the degree of bending of a sequence of points is determined by determining, as the degree of bending, the cosine of the angle between two straight lines passing two points separated by a predetermined number of points before and after each point included in the sequence of points. Is evaluated, and when the degree of bending in the vicinity of each point is maximized, it is determined that the bending point corresponds to the bending portion. Therefore, there is an advantage that an appropriate bending point accurately reflecting the shape of the object can be obtained.

According to the second aspect of the present invention, a threshold value is set for the degree of bending, and if the degree of bending is smaller than the threshold value, it is not adopted as a bending point, so that a minute fluctuation caused by a measurement error or the like can be removed. Therefore, there is an advantage that erroneous detection due to the influence of the measurement error can be suppressed. According to the method of claims 3 to 5, unnecessary bending points are removed after the bending point is obtained, so that there is an advantage that a more appropriate bending point can be obtained.

In particular, when the original point sequence has many straight lines,
By removing the error component based on the threshold value as in claim 2, the number of bending points to be evaluated as to whether or not it is unnecessary is greatly reduced, so that unnecessary bending as in claims 3 to 5 is achieved. Even if processing for removing points is added, there is an advantage that the amount of calculation is reduced as compared with the conventional division / synthesis method.

[Brief description of the drawings]

FIG. 1 is an operation explanatory diagram showing an embodiment.

FIG. 2 is a diagram showing an example of a point sequence in the embodiment.

FIG. 3 is a diagram illustrating a concept of a degree of bending in the embodiment.

FIG. 4 is a diagram illustrating a concept of a bending point in the embodiment.

FIG. 5 is an operation explanatory diagram of the second embodiment.

6A is a diagram illustrating an example of a bending point processed in the second embodiment, and FIG. 6B is an enlarged view of a main part of FIG.

FIG. 7 is a diagram illustrating a processing result according to the second embodiment.

FIG. 8 is an operation explanatory diagram of the third embodiment.

9A is a diagram illustrating an example of a bending point processed in a third embodiment, and FIG. 9B is an enlarged view of a main part of FIG. 9A.

FIG. 10 is a diagram illustrating a processing result according to the third embodiment.

FIG. 11 is an operation explanatory diagram of the fourth embodiment.

12A is a diagram illustrating an example of a bending point processed in a fourth embodiment, and FIG. 12B is an enlarged view of a main part of FIG.

FIG. 13 is a diagram illustrating a processing result according to the fourth embodiment.

FIG. 14 is a flowchart showing an example of a procedure for measuring a shape of an object using a bending point obtained according to the present invention.

FIG. 15 is a diagram showing a concept when measuring the shape of an object using a bending point obtained by the present invention.

FIG. 16 is a schematic configuration diagram of an apparatus used in an example.

FIG. 17 is a diagram showing a process of obtaining a point sequence by the device shown in FIG. 16;

FIG. 18 is a diagram showing the concept of a polygonal line approximation process in a conventional example.

FIG. 19 is a diagram showing the concept of a broken line approximation process in another conventional example.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 object 2 light source 3 imaging device 11 binarization processing unit 12 noise removal unit 13 skeleton extraction unit 14 storage unit 20 broken line processing unit 21 bending degree calculation unit 22 bending degree extraction unit 23 threshold processing unit 24 unnecessary point removal unit

Continuation of the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G01B 11/00-11/30 102 G01B 21/00-21/32 G06T 5/30 G06T 7/00

Claims (5)

(57) [Claims] 1. A method according to claim 1, wherein the position of a plurality of measurement points of an object having a bent portion on a line intersecting with a cutting plane traversing the bent portion is determined based on a point sequence in a predetermined order represented by two-dimensional coordinates. In the shape estimation method of approximating the shape of the object with a polygonal line, a cosine of an angle between two straight lines passing through two points separated by a predetermined number of points before and after each point forming a point sequence is obtained as a degree of bending, A shape estimating method characterized by determining a point where the degree of curvature is maximum in a vicinity range set for each point as a bending point, and obtaining a polygonal line composed of line segments connecting the bending points in order. 2. A method according to claim 1, wherein, among the points included in the point sequence, only a point where the degree of curvature exceeds a predetermined threshold value is determined as a point having a maximum degree of curvature in a range near the point. 1. The shape estimation method according to 1. 3. After obtaining a line segment connecting the inflection points, when the length of the line segment connecting the adjacent inflection points is shorter than a predetermined lower limit, at least one of the inflection points at both ends of the line segment 2. The shape estimating method according to claim 1, wherein one of them is excluded as an unnecessary bending point. 4. After calculating a line segment connecting the inflection points, when the angle between two line segments passing through the inflection point is larger than a predetermined upper limit, the inflection point is excluded as an unnecessary inflection point. The method according to claim 1, wherein: 5. A distance between a line segment connecting two adjacent bending points on both front and rear sides of the target bending point and a target bending point after obtaining the bending point, and the distance is defined as a predetermined threshold value. 2. The shape estimating method according to claim 1, wherein the target bending point is excluded as an unnecessary bending point when smaller than the target bending point.