JPH07111331B2 - 3D shape input device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a three-dimensional shape input device for inputting a three-dimensional shape of an object to be measured into a three-dimensional CAD system, computer graphics system or the like at high speed and with high accuracy.

[0002]

2. Description of the Related Art In recent years, three-dimensional CAD has been introduced into various industrial design fields as a tool for designing a three-dimensional model. However, direct modeling with three-dimensional CAD requires considerable training due to the difficulty of shape input. Therefore, the designer creates a design model made of clay or wood, measures it, and CAD
Is used, and the measurement of the design model is performed manually, or by a contact type or non-contact type three-dimensional digitizer. As a conventional three-dimensional shape input device, for example, a structure described in Nikkei Computer Graphics, September 1988, "Revolution in CAD data input or simplified three-dimensional shape input device" is known.

A conventional three-dimensional shape input device will be described below. FIG. 48 shows a conventional three-dimensional shape input device. In FIG. 48, 4801 is a laser light source for generating slit light for irradiating an object to be measured, 4802 is a mirror for changing the optical path of the slit light, 4803 is slit light, 4804 is an object to be measured, 4805 is an object to be measured 480.
X-axis moving mechanism for moving 4 4806 for slit light 4
A camera for capturing the scattered light of 803, a camera 4807
A / D converter for converting the output signal of 806 into a digitized image signal, 4808 is an image memory for storing the image signal, and 4809 is a slit scattered light center position for calculating the center position of the slit scattered light by the image signal. Detecting means, 4810 is a coordinate calculating means for calculating a three-dimensional coordinate value from the center position of the slit scattered light, 4811 is a data format converting means for converting the measured three-dimensional coordinate value data into a data format for the input destination system, 4812.
Is a scanner control means for controlling the entire system, 4813 is three-dimensional shape data obtained by this apparatus, and 4814 is three-dimensional CA.
It is a D system.

The operation of the three-dimensional shape input device configured as described above will be described below. First, the slit light 4803 from the laser light source 4801 is reflected by the mirror 4
The optical path is changed by 802, and the measured object 4804 is irradiated. The DUT 4804 moves in the X-axis direction at a constant pitch by the X-axis moving mechanism 4805 while receiving the slit light 4803. The camera 4806 is a moving object under measurement 4804.
The slit scattered light from is imaged in synchronization with the movement. In this case, the captured image of the slit scattered light on the camera 4806 shows unevenness of the slit scattered light according to the shape of the surface of the DUT 4804. The A / D converter 4807 converts the image analog signal of the camera 4806 into a digital signal, and this signal is temporarily stored in the image memory 4808 as an image signal. Slit scattered light center position detecting means 4
Reference numeral 809 obtains the center position of the brightness of the slit scattered light for each horizontal scanning line of the camera from the image signal and regards it as the image of the measurement point. In the coordinate calculation means 4810, the angle formed by the straight line connecting the image of the measurement point and the lens center of the camera and the optical axis of the camera, the baseline length (distance from the slit light source to the lens center of the camera), and the projection angle of the laser slit light are shown. The three-dimensional coordinate value of the measuring point is calculated by using the principle of triangulation method. In this way, the three-dimensional shape data of the measured object is acquired for one slit light. And X-axis moving mechanism 4805
By acquiring 3D shape data for each movement pitch of
Finally, the three-dimensional shape data of the entire measured object is acquired.
The data format conversion means 4811 measures the object under measurement 480.
The shape data of the entire three-dimensional coordinate value of 4 is converted into a data format for the input destination system as point sequence data or mesh-like data, and the format is converted.
The dimensional shape data 4813 is a three-dimensional CAD system 481.
4 is input and displayed.

[0005]

However, in the above-mentioned conventional configuration, the amount of data is enormous because the number of points to be measured is large, so that redrawing, rendering, etc. by changing the viewpoint due to the memory relationship of the three-dimensional CAD system. However, there is a problem in that the image generation function of does take a lot of time and the operability is deteriorated. Moreover, since the input three-dimensional shape data is a point sequence, there is a problem that it is very difficult to change and correct the shape in the system. In particular, the rounded portion is a portion where a request for correction is relatively likely to occur when the design stage shifts to the design stage, and it is necessary to input data to the CAD system in a data format that can cope with such a correction. In particular, in the case of design clay mockup, it is desirable to be able to adjust these fine parts in CAD, because it is difficult and time-consuming to create accurate are and inverse are on clay. Also, if you want to partially use another model, if you can take out the data of that part and combine it with other parts to create a new model, the efficiency of the design process will improve, but in reality Since the rounding transformation operation is often performed, the boundary of the partial data needs to be a data format that can be easily rounded.
However, these rounding deformation operations are quite difficult with point sequence data.

The present invention solves the above-mentioned problems of the prior art. Firstly, the three-dimensional shape of an object to be measured is measured at high speed and with high accuracy, and the measured shape is saved, and the amount of efficient data is reduced. The data is smoothed, and secondly, it has a data format that makes it easy to change and correct the shape in the input destination system.
It is an object to provide a three-dimensional shape input device for inputting into a three-dimensional CAD system or a computer graphics system.

[0007]

In order to achieve this object, the present invention firstly provides a slit light source for generating slit light for irradiating an object to be measured, a moving mechanism for moving the object to be measured, and an object to be measured. A camera for capturing scattered light of slit light from an object, an A / D converter for converting a luminance signal from the camera into a digitized image signal, an image memory for storing the image signal, and an image memory Slit scattered light center position detecting means for detecting the center position of the slit scattered light from the stored image signal, coordinate calculating means for calculating a three-dimensional coordinate value from the slit scattered light center position, and indicating an edge on the object to be measured. And the parameters of the parametric curved surface are u and v, the parameters of the measurement points on the continuous edges are set to the parameter values u or v so that they are aligned. And parameter setting means for applying a data, node and control points to obtain an approximate error following function specified three-dimensional coordinate value data representing the entire shape of the workpiece
When the increasing and moving are repeated and the three-dimensional coordinate value data is approximated to a parametric curved surface, the increasing direction and the increasing position of the nodes and control points are automatically determined. For each set U _i of measurement point sequences having the same u value and set V _j of measurement point sequences having the same parameter v value, the variance Uvar of the difference in depth coordinate values between the previous approximation and the three-dimensional coordinate value data. _i (i = 1, ...
.., m), Vvar _j (j = 1, ..., N) are calculated, the averages U and V of the variances Uvar _i and Vvar _j are obtained, and the magnitudes are compared. If U is large, V direction, V
Is larger, it is decided to increase the number of nodes and control points in the u direction. For example, Uvar _i
At (i = 1, ..., M), a set U ₀ of measurement point sequences having the largest variance value is obtained, and the sum of squares of the error of the point sequence U ₀ is calculated for each node to obtain the most square. A new node is added to the center between nodes with a large sum value.If the approximation error decreases at a low rate even if the number of control points is increased, the surface of the DUT is divided into multiple parts in the direction where the fluctuation of the error is large. However, if the approximation error decreases considerably even if the control points are increased by performing parametric function approximation as different curved surfaces, polygon patches are created using the feature points of the measured three-dimensional coordinate value data and the surface is created. And a data format conversion means for converting the data compressed by the data compression means into the data format of the input destination system.

Secondly, a slit light source for generating slit light for irradiating the object to be measured, a moving mechanism for moving the object to be measured, a camera for capturing scattered light of the slit light from the object to be measured, and the camera A / D converter for converting the luminance signal of the above into an image signal digitized, an image memory for storing the image signal, and slit scattering for detecting the center position of the slit scattered light from the image signal stored in the image memory If the light center position detecting means, the coordinate calculating means for calculating a three-dimensional coordinate value from the slit scattered light center position, the curve indicating means for indicating the edge on the object to be measured, and the parameter of the parametric curve are u, the parameters are continuous. Parameter setting means for assigning a parameter to each measurement point so that the values of the parameter u of the measurement point on the edge are aligned, and a cubic representing the overall shape of the measured object. When approximating the three-dimensional coordinate value data repeatedly increased and movement of nodes and control points to obtain an approximate error following function specified coordinate value data to a plurality of parametric curves, automatically increasing position of nodes and control points As a method for determining the total, the sum of squares of the approximation error of the point sequence U that approximates one parametric curve is calculated for each node,
A new node is added to the center between the nodes with the largest sum of squares. If the rate of decrease in the approximation error is low even if the number of control points is increased, the curve is divided into multiple parametric curves. A data compression unit that performs approximation and then interpolates a plurality of parametric curves to create a curved surface, and a data format conversion unit that converts the data compressed by the data compression unit into the data format of the input destination system. Have a configuration.

Thirdly, in addition to the above-mentioned first and second components, a curve designating means for determining a curve for dividing the object to be measured into a plurality of areas, and the curve to divide the object to be measured into a plurality of areas. And a data compression means for approximating the three-dimensional coordinate value data to a parametric function for each area. However, the curve designating means used at the time of parameter setting and area division specifies the curve on the object to be measured using a color whose reflectance to the slit light is different from the surface color of the object to be measured, and the brightness value in the image signal. Is extracted as the curve, or the reflectance to the slit light is specified by using a color different from the surface color of the DUT to specify the rough curve on the DUT and the brightness value in the image signal. Is extracted based on the edge-likeliness obtained using the three-dimensional coordinate value data from the different portion and its peripheral portion, or the obtained curve is regarded as the edge portion of the object to be measured, and human intervention is required. Without the need to automatically obtain the edge using the three-dimensional coordinate value data, or the camera reads the slit scattered light and does not project the slit light. A memory switching unit for capturing an image of the whole and switching the output depending on whether the camera captures the slit light or the luminance information of the entire object to be measured which does not project the slit light, and the memory switching unit. A first image memory for storing the image signal of the slit scattered light by switching the above, and a second image memory for storing the entire brightness image of the measured object not projecting the slit light by switching the memory switching means.
Image memory is separately provided, and the curve to be obtained is regarded as the edge portion of the object to be measured, and the edge is automatically obtained by using the brightness image without human intervention, or the surface color of the object to be measured is determined. A color camera that specifies the curve on the object to be measured using a color different from the above, reads the slit scattered light by the camera, and captures the entire object to be measured in the state where the slit light is not projected. Memory switching means for switching the output depending on whether the image is captured or the brightness information of the entire object to be measured which is not projecting the slit light, and the image signal of the slit scattered light is stored by switching the memory switching means. A first image memory and a second image for storing a brightness image of the entire object to be measured which is not projecting slit light by switching the memory switching means. Separately from the brightness image, the curve is extracted from the brightness image using the difference in color, or the measured three-dimensional coordinate value data is displayed on a computer, and the mouse is moved to the displayed three-dimensional coordinate value data. It is used by humans to indicate curves.

Fourthly, in addition to the above-mentioned third constituent element, boundary discrimination for discriminating whether the boundary portion of the area divided by the area dividing means is a roof edge, a smooth edge of a convex curved surface or a smooth edge of a concave curved surface. And a rounding radius estimating means for estimating rounding radii of the convex curved surface and the concave curved surface, and an area boundary generating means for regenerating a special area boundary in the convex curved surface and the concave curved surface. Further, in addition to the above-mentioned constituent elements, it has a configuration of area combining means for combining areas by performing a rounding deformation operation based on the rounding radius estimated by the rounding radius estimating means.

[0011]

According to the present invention, according to the above configuration, firstly, the three-dimensional coordinate value data representing the entire shape of the object to be measured is approximated to a parametric function having a predetermined approximation error or less with a data amount as small as possible. It is possible to efficiently reduce the amount of data in which the saved shape is saved and smooth the data.

Secondly, by dividing the three-dimensional coordinate value data of the object to be measured into main regions and performing a function approximation, and further estimating the boundary shape and the rounding radius of each area to create a special boundary shape. It is possible to input to a three-dimensional CAD system or a computer graphics system in a data format that can easily change and correct the shape in the input destination system.

Thirdly, by setting the parameter values so as to be aligned along the edges when performing the parametric function approximation, it is possible to reduce the shape of the undulation that normally occurs when performing the curved surface approximation.

Fourth, when performing approximation to a parametric curved surface, iterative approximation is repeated in which the number of nodes and control points is increased, and the direction of increasing control points and the position of increasing nodes are determined using the variance of the previous approximation error. By deciding, the problem of the number of control points and the position of nodes to obtain more efficient approximation, which was difficult to predict directly from the shape of the object to be measured, was solved, and the three-dimensional coordinate value data was reduced as much as possible. A quantity can be approximated to a parametric surface.

[0015]

(Embodiment 1) A first embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block connection diagram of a three-dimensional shape input device according to a first embodiment of the present invention.

In FIG. 1, 101 is a laser light source for generating slit light for irradiating an object to be measured, 102 is a mirror for changing the optical path of the slit light, 103 is slit light for irradiating the object to be measured, and 104 is an object to be measured. Reference numeral 105 denotes an X-axis moving mechanism that moves the DUT 104 in the X-axis direction at a constant pitch. Reference numeral 106 denotes a camera that reads scattered light of the slit light 103. Reference numeral 107 converts an output signal of the camera 106 into a digitized image signal. A / D converter, 108 is an image memory for storing an image signal, 109 is a slit scattered light center position detecting means for calculating the center position of slit scattered light from the image signal, and 110 is a three-dimensional coordinate from the center position of the slit scattered light. Coordinate calculation means for calculating a value, 111 is a curve designating means for designating a curve on the object to be measured, and 112 is a parameter of a parametric curved surface. Where u and v are parameter setting means for assigning a parameter to each measurement point so that the values u or v of the parameter values of the measurement points on the continuous curve are aligned, 113 denotes the overall shape of the object to be measured. Parametric curved surface approximating means for approximating the three-dimensional coordinate value data to the parametric curved surface, and 114 for the parametric curved surface approximating means 11.
Data format conversion means for converting the data compressed by 3 into the data format of the input destination system, 1
Reference numeral 15 is a scanner control means for controlling the entire system, 116 is three-dimensional shape data obtained by this apparatus, 117 is a three-dimensional CA.
It is a D system.

Instead of the X-axis moving mechanism 105 that moves the object 104 to be measured, slit light is emitted from the object 104 to be measured.
It may be a slit light scanning means for scanning above.

The operation of the three-dimensional shape input device configured as described above will be described. First, a person fills in a curve used for setting parameters on the object to be measured 104 by using a color whose reflectance to slit light is higher than the surface color of the object to be measured 104. The color to be used may be any color as long as the reflectance with respect to the slit light can be differentiated from the surface color of the DUT 104. This operation is 2 when the curved surface is approximated.
Since it has the effect of reducing the waviness shape that normally occurs in the vicinity of an edge that is oblique in two parameter directions, it is advisable that the indicated curve traces an oblique edge in two parameter directions. Slit light 103 from laser light source 101
The light path is changed by the mirror 102 and the object to be measured 104 is irradiated. The DUT 104 is moved in the X-axis direction at a constant pitch by the X-axis moving mechanism 105 while receiving the slit light 103. The camera 106 images the slit scattered light from the moving DUT 104 in synchronization with the movement. In this case, the captured image of the slit scattered light on the camera 106 shows the unevenness of the slit scattered light according to the shape of the surface of the DUT 104. The A / D converter 107 is the camera 106.
Signal is converted into a digital signal, and this signal is temporarily stored in the image memory 108 as an image signal. The slit scattered light center position detecting means 109 finds the center position of the luminance of the slit scattered light for each horizontal scanning line of the camera from the image signal and regards it as the image of the measurement point. In the coordinate calculation means 110, the angle formed by the straight line connecting the image of the measurement point and the lens center of the camera and the optical axis of the camera, the baseline length (distance from the slit light source to the lens center of the camera), and the projection angle of the laser slit light are shown. The three-dimensional coordinate value of the measuring point is calculated by using the principle of triangulation method. In this way, the three-dimensional shape data of the measured object is acquired for one slit light. Then, by obtaining the three-dimensional shape data for each moving pitch, the three-dimensional shape data of the entire object 104 to be measured is finally obtained. The curve designating means 111 recognizes the position of the curve written on the DUT 104. Parameter setting means 1
If the parameters of the parametric curved surface are u and v, 12 assigns parameters to the respective measurement points so that the values u or v of the parameter values of the measurement points on the continuous curve are aligned. The parametric curved surface approximating means 113 compresses the shape information by approximating the three-dimensional coordinate values to a parametric curved surface. The data format conversion means 114 converts the parameter values of a plurality of parametric curved surfaces representing the shape of the entire measured object approximated by the parametric curved surface approximating means 113 into the data format for the CAD system of the input destination, and the format-converted three-dimensional. The shape data 116 is input to the three-dimensional CAD system 117.

The detailed operation of the curve designating means 111 will be described below. First, the maximum luminance value is obtained for each horizontal scanning line of the camera in the image temporarily stored in the image memory 108, and this is set as the luminance value at the measurement point. By repeating this for each image, the luminance value at each measurement point is obtained. A measurement point whose luminance value at the measurement point is larger than a predetermined threshold value is obtained. By tracing these measurement points, the designated curve on the DUT 104 is extracted. The extracted curve is subjected to expansion processing and thinning processing to create a continuous curve with one width.

The curve designating means uses the color whose reflectance to the slit light is different from the surface color of the object to be measured to specify the rough curve on the object to be measured, and the brightness value is different in the image signal. The curve is extracted from the portion and the peripheral portion thereof based on the edge-likeness obtained using the three-dimensional coordinate value data.

Alternatively, the curve to be obtained is regarded as the edge portion of the object to be measured, and the edge is automatically obtained using the three-dimensional coordinate value data without manual intervention.

Alternatively, the camera reads the slit scattered light and images the entire object to be measured in a state where the slit light is not projected, and the camera either captures the slit light or does not project the slit light. A memory switching unit that switches output depending on whether or not the brightness information of the entire object to be measured is imaged, a first image memory that stores an image signal of slit scattered light by switching the memory switching unit, and a memory switching unit that switches the memory switching unit. A second image memory for storing the whole luminance image of the object to be measured which is not projecting the slit light is separately provided, and the obtained curve is regarded as an edge portion of the object to be measured, without human intervention. An edge is automatically obtained using the luminance image.

Alternatively, the curve is designated on the object to be measured using a color different from the surface color of the object to be measured, and the camera reads the slit scattered light and images the entire object to be measured without projecting the slit light. In the color camera, the output is switched depending on whether the camera images the slit light or the luminance information of the entire object to be measured which does not project the slit light. A first image memory for storing the image signal of the slit scattered light and a second image memory for storing the brightness image of the entire object to be measured which is not projecting the slit light by switching the memory switching means are separately provided. The curve is extracted from the luminance image by using the difference in color.

Alternatively, the measured three-dimensional coordinate value data may be displayed on a computer, and a human may use the mouse to indicate a curve for the displayed three-dimensional coordinate value data.

The parameter setting means 112 described above will be described below.
The more detailed operation will be described. With respect to a continuous curve having a width of 1 point extracted by the curve designating means 111, first, one curve has one constituent point per slit and both end points of the curve are the first slit and the final slit. Rearrange the curves to fit. FIG. 2 is an example of a curve in which one curve has one constituent point for each slit and both end points of the curve are on the first slit and the final slit. Reorganization is performed as follows. When one curve does not have one point for each slit, one edge is divided into a plurality of curves so that one edge has one point for each slit. FIG. 3A shows an example in which the number of constituent points of one curve is not one per slit, and FIG.
This is an example of division into a plurality of curves so that one edge has one constituent point per slit. Then, by extending the curve from each end point of the curve in parallel to the scanning direction of the slit, knitting is performed so that both end points of the curve are placed on the first slit and the final slit. Similar processing is performed when the curve has one point for each slit and both end points of the curve are not on the first slit and the final slit. FIG. 4A shows an example of a case where the curve has one point for each slit and both end points of the curve are not placed on the first slit and the final slit, and FIG. 4B shows the curve. This is an example in which a curve is knitted so that both end points are placed in the first slit and the final slit. Next, the scanning direction of the slit is set as the parameter u direction, and the value of the parameter u at each point is determined. Since the interval between the slits is constant, the value of the parameter u at each point is set as follows:
It is sufficient to divide 0 to 1 evenly by all slits from the first slit to the final slit. The values of the parameter u at all points on one slit are the same. And the parameter u
The value of the parameter v is obtained by v in the direction orthogonal to. The value range of the parameter v is also 0 to 1. First, consider one slit, which is the center of the first slit and the final slit. This slit is called the central slit. For the data of the central slit, the value of the parameter v is determined using (Equation 1).

[0027]

[Equation 1]

In (Equation 1), data is a three-dimensional array in which measured three-dimensional coordinate value data is stored, and d
It is defined by ata [snum] [dnum] [3].
snum is the number of slits, and dnum is 3 in each slit.
It is the number of dimensional coordinate value data, and 3 represents three-dimensional coordinate values of x, y, and z. Further, Mid_Slit represents the central slit, and Mnum represents the number of three-dimensional coordinate value data in the central slit. Next, the value of the parameter v of the remaining slits other than the central slit is determined. FIG. 5 is an explanatory diagram for setting the value of the parameter v by the designated curve. The curves obtained by organizing the curves obtained by the curve designating means 111 are arranged in order from the one having the value of the parameter v closer to 0. First, the value v _mid of the parameter v of the data point corresponding to the constituent point of the first curve in the central slit is taken out. All the values of the parameter v of the data points corresponding to the constituent points of the first curve at the other slits are set to v _mid . By performing this operation on all curves, the value of the parameter v of the data points on the curves is determined. Then, in each slit, the value of the parameter v of the data point between the curves is linearly given according to the y coordinate value of the data point. FIG. 6 is an explanatory diagram for setting the value of the parameter v in the remaining data points of each slit. By the above operation, the parameter v of the data point
The values of are set to have the same value on successive curves. By setting this parameter, the poor approximation of the waviness shape that usually occurs during curved surface approximation is reduced.

Hereinafter, the above-mentioned parametric surface approximation procedure
A more detailed operation of stage 113 will be described. Figure 7 shows the parameters
3D measured by the trick curved surface approximating means 113
The coordinate data is Non, which is one of the parametric curved surfaces.
-Uniform B-Spline (Non-Uniform
BM spline) Approximate accuracy specified for curved surface or less
Shows the flow of the method approximated by. Represent a curved surface
The parameters are u and v. First, measure in step 701
The three-dimensional coordinate value of the measured object is input. In step 702
Number of control points in the u direction (n_u+1) and floor m_uIs specified. hand
In order 703, the number of control points and the rank of curved surface m in the number of nodes in the u direction_u
The value with is added. Nodal position in u direction in step 704
Set the positions at regular intervals. However, the nodes at both ends are m_u
Multiple times. In step 705, the number of control points in the v direction (n_v+
1) and the number of floors m_vIs specified. Nodes in v direction in step 706
Number of control points and surface rank m_vIs added to
It In step 707, set the v-direction node positions at equal intervals.
It However, the nodes at both ends are m _vMultiple times. procedure
In 708, the nodes obtained in steps 704 and 707 are
By using the data and applying the least squares approximation to (Equation 2),
Calculate control points. P in (Equation 2)_ijIs the control point, N
_{i, p}, N_{j, q}Is the spline basis function.

[0030]

[Equation 2]

In step 709, the approximation error of the least square approximation performed in step 708 is calculated. In step 710, it is determined whether the approximation error is less than the specified allowable error. If the approximation error is less than the specified allowable error, the procedure proceeds to step 711, and in step 711, the calculated control points in the u and v directions,
And the coordinate values of the nodes are output as parameters representing the shape. If the allowable error is not satisfied in step 710, the number of nodes and the number of control points are increased by the same number in step 712.
Then, returning to step 707, the least square approximation is performed again. By repeating the above procedure, an approximate curved surface that satisfies the specified tolerance can be realized with a small amount of data.

As described above, according to this embodiment, the laser light source 101 and the mirror 102 as the slit light source for generating the laser slit light for irradiating the object to be measured 104, and the object to be measured 104 at a constant pitch in the X-axis direction. An X-axis moving mechanism 105 for moving the object, a camera 106 for imaging the scattered light of the slit light by the DUT 104, an A / D converter 107 for converting an output signal from the camera 106 into a digitized image signal, Image memory 10 for storing the image signal
8, a slit scattered light center position detecting means 109 for detecting the center position of the slit scattered light from the image signal stored in the image memory 108, and a three-dimensional coordinate value of the DUT 104 from the slit scattered light center position. Coordinate calculation means 11
0 and a curve designating means 111 for designating a curve on the object to be measured.
And the parameters of the parametric curved surface are u and v, the parameter setting means 112 for assigning a parameter to each measurement point so as to make the parameter values u or v of the measurement points on the continuous curve uniform. Parametric curved surface approximating means 113 for compressing shape information by approximating three-dimensional coordinate value data representing the overall shape of the measured object to a parametric curved surface, and the parametric curved surface approximating means 11
Data format conversion means 1 for converting the parametric curved surface representing the overall shape of the DUT approximated by 3 into the data format for the CAD system 117 of the input destination.
14 and the scanner control means 115 for controlling the entire system, the amount of data is efficiently reduced while the measured shape is saved, and when the curved surface is approximated, the vicinity of the diagonal edge in the directions of the two parameters is reduced. It is possible to reduce the undulating shape that often occurs and to input it to a three-dimensional CAD or computer graphics system in a data format in which the shape can be easily changed or modified in the input destination system.

(Second Embodiment) A second embodiment of the present invention will be described below with reference to the drawings.

FIG. 8 is a block diagram of a three-dimensional shape input device according to the second embodiment of the present invention. In FIG. 8, 801
Reference numeral 802 is a laser light source for generating slit light for irradiating the object to be measured, 802 is a mirror for changing the optical path of the slit light, 8
Reference numeral 03 denotes slit light irradiating the measured object, 804 denotes the measured object, 805 denotes an X-axis moving mechanism for moving the measured object 804 at a constant pitch in the X-axis direction, and 806 a camera for reading scattered light of the slit light 803, Reference numeral 807 denotes an A / D converter for converting the output signal of the camera 806 into a digitized image signal, 808 an image memory for storing the image signal, and 809 a slit scattered light center for calculating the central position of the slit scattered light by the image signal. A position detecting means, 810 coordinate calculating means for calculating a three-dimensional coordinate value from the center position of the slit scattered light, 811 a curve designating means for designating a curve on the object to be measured, 812 a u, v parameter of the parametric curved surface.
Then, the parameter setting means 814 assigns a parameter to each measurement point so that the values u or v of the parameter values of the continuous measurement points on the curve are aligned, and 814 indicates the data compressed by the parametric curved surface approximation means 813. Data format conversion means for converting to the data format of the input destination system, 815 is scanner control means for controlling the entire system, 816 is three-dimensional shape data obtained by this device, 8
Reference numeral 17 is a three-dimensional CAD system, which has the same configuration as that shown in FIG.

The structure of FIG. 1 is different from that of FIG.
The lametric curved surface approximating means 113 is compared with the measured surface in FIG.
3D coordinate value data representing the entire shape of the fixed object 804 is a node
Parameter and the number of control points
How to increase the number of nodes and control points when approximating to a rick surface
Defining nodes as a way to automatically determine orientation and position
If the directions are u and v, the values of parameter u are equal.
Set U of fixed points_i, And the measurement with the same value of parameter v
Set of point sequence V_jEach time, the previous approximation and the three-dimensional coordinate value
Variance Uvar of the depth coordinate value difference with the data_i(I = 1,
..., m), Vvar_jCalculate (j = 1, ..., n)
Out, and then the distributed Uvar_i, Vvar _jMean U, V
, And compare the magnitude, if U is large, in the v direction,
If V is large, increase the number of nodes and control points in the u direction.
As the variance of the larger mean, for example Uvar
_iThe variance with the largest value at (i = 1, ..., m)
Set U of measurement point sequences with values₀To obtain the point sequence U₀Error of
The sum of squares is calculated for each node, and the value of the sum of squares is the largest.
Automatic node determination means 8 for increasing a new node at the center between points
18 is provided, and the parametric curved surface approximating means 813 is a node.
Non-uniform determined by the automatic determination unit 818
Parametric data of 3D coordinate values using different nodes
It is a point that is approximated to a curved surface by least squares.

The operation of the three-dimensional shape input device configured as described above will be described. First, the DUT 80
The procedure is the same as that of the first embodiment until the entire three-dimensional shape data of No. 4 is acquired and the parameters are set at each data point measured by the parameter setting unit 811. Next, the automatic node determination means 818 automatically determines the increasing directions and positions of appropriate nodes and control points, and the parametric curved surface approximating means 813 approximates the three-dimensional coordinate values to the parametric curved surface using the determined nodes. To do. Data format conversion means 81
4 converts the data compressed by the parametric curved surface approximating means 813 into the data format of the input destination system, and the format-converted three-dimensional shape data 816 is input to the three-dimensional CAD system 817.

Hereinafter, the parametric curved surface approximating means 813
A more detailed operation of the above will be described. Parameters representing a curved surface are u and v. First, the three-dimensional coordinate value of the data point having the parameter set by the parameter setting means 811 is B-Spli at an appropriate number of control points corresponding to the number of data points.
ne The first least-squares approximation is performed on the curved surface. At this time, the node is a uniform. The parametric curved surface approximating means 813 repeats curved surface approximation to approximate the measured shape with a relatively small number of control points. Therefore, it is necessary to set the end condition for the approximation repetition. The termination condition is set as follows. FIG. 9 is a flow chart for explaining the termination condition. First, three-dimensional coordinate values of points on the B-Spline curved surface are calculated by substituting the parameter values of the data points into the B-Spline curved surface having the control points obtained as a result of the least square approximation. There are two evaluation criteria. In criterion 1, the calculated x,
The square of the residual between the y coordinate value and the x and y coordinate values of the measurement data point is calculated for each point. In Criterion 2, the sum of squares of the residual of the z coordinate value is calculated. Divide it by the number of data and calculate the root mean square of the residuals. As described above, the criteria 1 and criteria 2 are calculated, and the evaluation is performed as follows. Even if only one of the x- and y-coordinate residuals of each point in the reference 1 is one, the predetermined thresholds Errorx and Error are set.
If it exceeds, it is determined that the approximation is bad, a new node is calculated in 904, and the next approximation is performed in 901. On the other hand, if the x and y coordinate value residual values are small at all points, the criterion 2
Leave it to the decision. Criterion 2 is to calculate a new node at 904 when the value of the residual root mean square is smaller than the value of the previous approximation result, and perform the next approximation at 901. The approximation is terminated with the best approximation. As described above, the reason why the different evaluation operations are performed by dividing the approximation evaluation standard into the standard 1 and the standard 2 is that the residual of the standard 1 in the x and y directions is approximate and the shape created by the original data. This is because there is a considerable visual difference between the two. A large residual in the z direction means that the smoothness of the shape is high, but a residual in the x and y directions reminds us of a shape that is completely different from the original shape. FIG. 10 is a diagram for explaining a difference in appearance of residuals of x and y coordinate values and residuals of z coordinate values. When FIG. 10A is the original data obtained by measuring the human face, the approximation result of this data shows that when the residual of the x and y coordinate values is large, the residual of the z coordinate value of FIG. Is large and the residual of the x and y coordinate values is small, the result is as shown in FIG. Next, the second and subsequent approximations will be described. Two
In the subsequent approximations, the result of the previous approximation is always used to determine the position at which a new node is to be set from the difference from the measured data, and the least square approximation of the data point is performed again to calculate the control point position. At this time, from the relational expression between the nodes and the number of control points,
The number of control points increases in the same direction as the number of nodes increases. The calculation of a new node position is performed by the automatic node determination means 818.

The more detailed operation of the automatic node determining means 818 will be described below. First, the slit having the smallest number of data points of all the slits is obtained and referred to as the minimum size slit. Then, the value of the parameter v of the minimum size slit is used as the reference parameter in the v direction. Then, a parameter formed in a grid pattern on the uv plane using the reference parameter in the v direction will be referred to as a reference grid parameter. Then, the coordinate value of the point on the approximated curved surface in the reference grid parameter is obtained. Next, the three-dimensional coordinate value of the measurement data point in the reference grid parameter is temporarily calculated.
Since the slits are arranged at equal intervals, the parameter u of the data point is the same for each slit, but the value of the parameter v is different for each slit, so the temporary position of the measurement data point at the grid-like parameter position is set. Is calculated by linearly interpolating from the position of the original measurement data point using (Equation 3).

[0039]

[Equation 3]

(Equation 3) is an expression for obtaining the coordinate value of the temporary measurement data point in the parameter value (u, v).
ter_v [i] [j-1] ≤ v <parameter_v [i] [j]
i, j are calculated and the coordinate values are calculated using (Equation 3). That is, i is determined by the parameter u. However, (Equation 3)
, W is the value of the parameter v of the reference lattice, i is the slit number, j is the data number in each slit, parame
ter_w [i] [j] is the value of the parameter v at the jth measurement data point in the ith slit, and ratio is parameter_
The position of parameter v, where v [i] [j-1] is 0 and parameter_v [i] [j] is 1, data [i] [j] [1] is at the i-th slit The y coordinate value of the jth measurement data point, data [i] [j] [2] is the z coordinate value of the jth measurement data point in the ith slit, and new_data [i] [j] [1] is i The y-coordinate value of the j-th tentative measurement data point in the n-th slit, new_data [i] [j] [2] is the z-coordinate value of the j-th tentative measurement data point in the i-th slit. Note that the parameter for the smallest size slit was used to create the reference grid because the data is dense with respect to the number of parameters on the slit with many measurement data points, so provisional measurement data on the slit with many measurement data points is used. This is because the interpolation of the position of the point gives a better interpolation result than the interpolation on the slit having few measurement data points. Also,
Even if the approximate evaluation is performed using the residuals of the points on the curved surface with the reference grid parameter and the temporary measurement points, there is no difference in the evaluation results although there is a numerical difference. The approximation termination condition is that the residual root mean square value decreases as the number of nodes increases, and the previous approximation is regarded as the best approximation when the residual root mean square value increases. However, the decrease and increase of the residual mean square fluctuate in the same way. FIG. 11 is a diagram showing the relationship between the reference grid parameter and the parameter of the actual measurement data point. In FIG. 11, 1101 is a measurement data point, 1102 is a reference grid parameter, 1103 is a temporary measurement data point, and 1104 is a minimum size slit. Since the three-dimensional coordinate value of the temporary measurement data point has been obtained as described above, the difference between the point on the approximate curved surface and the temporary measurement data point in the grid-like parameter according to the reference parameter is then obtained. The difference between the approximation result and the measured data point is calculated as follows in consideration of the square of the residual with positive and negative signs. The z-coordinate value of the temporary measurement data point is subtracted from the z-coordinate value on the approximated curved surface, and the positive / negative value is stored. Then, the positive and negative signs stored in the square of the residual are added. The reason why the residual square value is obtained with a sign in this way is that even if the residual square value is the same, if the positive and negative values are different, it is visually felt that there is a large difference from the original data shape. On the contrary, if the positive and negative signs are the same,
I feel the difference is small. FIG. 12A is an example in which the absolute values of the residuals are the same in the neighborhood region but the signs are different, FIG.
Is an example when the absolute value and the sign of the residual are the same in the neighborhood area. In FIG. 12, 1201 is a surface formed by the measurement data, and 1202 is an approximate curved surface. The direction and position of setting a new node are determined as follows using the signed residual squares at each point of the reference lattice parameter calculated as described above. FIG. 13 is a diagram illustrating a set of measurement point sequences for calculating variance. First, as shown in FIG. 13, a set of measurement point sequences U _i (i
= 1, ..., m) Signed residual squared variance Uvar
_i (i = 1, ..., M) and variance Vvar _j (j) of signed residual squares of a set V _j (j = 1, ..., n) of measurement point sequences with the same value of parameter v = 1, ..., n)
Ask for. However, m is the number of all slits, and n is the number of reference parameters w. Then, Uvar _i (i = 1, ...
.., m) and Vvar _j (j = 1, ..., N) are averaged U and V, respectively, and the magnitudes are compared. If the average value U is larger, a new node is added in the v direction. This is because if the variance is large, nodes should be added to the places where the residual is large. FIG. 14 is an explanatory diagram of a direction of adding a node. In FIG. 14, 1401 is a slit having a large dispersion, 1402 is a slit having a small dispersion, 1403 is a position of a new node, 1404 is a curve on a surface formed by measurement data, and 1405 is a curve on an approximate curved surface. If the average value V is larger, a new node is added in the u direction. With the above, the directions for adding the nodes and the control points have been determined. Next, the position where a node is added will be described. For the sake of explanation, hereinafter, only the case where the average value U is larger will be described. The same explanation can be applied when the average value V is larger than U. FIG. 15 is an explanatory diagram of a position where a node is added. In FIG. 15, 1501 is the slit having the lowest degree of approximation, 1502 is the current node position, 1503 is the new node position, 1504 is the measurement data point, and 1505.
Is an approximate curved surface. First, the variance value Uvar _i (i = 1,
, M), the slit number i ₀ having the largest value is obtained. By this operation, the slit number with the lowest degree of approximation can be obtained. Next, the sum of squares of the residuals between the z-coordinate values of the points on the approximate curved surface and the z-coordinate values of the provisional measurement data points is calculated between the obtained nodes on the slit. Since there are a plurality of nodes, one residual sum of squares can be obtained for each node. Then, the largest value of the sum of squares of these residuals is obtained, and a new node is added to the center between the nodes. Even if there is a large residual squared value, this new method for automatically determining the nodal position is possible if the neighborhood has a similar residual squared value and is displaced in the same direction with respect to the measurement data points. , We use the variance of the signed residual squares based on the idea that the area does not feel a big difference in shape visually, and therefore it is not considered as a place where the approximation failure is large.

With the non-uniform nodes to which the new nodes obtained by the above operation are added, the measurement data is again B-
Spline Approximates a curved surface. The automatic node determining means 818 and the parametric curved surface approximating means 81
By repeating step 3, measurement data can be obtained with fewer control points.
It can be efficiently approximated to Non-Uniform B-Spline.

As described above, according to this embodiment, the laser light source 801 and the mirror 802, which serve as a slit light source for generating the laser slit light for irradiating the object to be measured, and the object 8 to be measured.
X-axis moving mechanism 805 that moves 04 in the X-axis direction at a constant pitch, camera 806 that captures the scattered light of the slit light by DUT 804, and the output signal from camera 806 is converted into a digitized image signal. A / D converter 80
7, an image memory 808 that stores the image signal, a slit scattered light center position detection unit 809 that detects the center position of the slit scattered light from the image signal stored in the image memory 808, and a slit scattered light center position detection unit 809. Measured object 80
4. If the coordinate calculation means 810 for calculating the three-dimensional coordinate value of 4, the curve indication means 811 for indicating the curve on the object to be measured, and the parameters of the parametric curved surface are u and v, the measurement points on the continuous curve are Parameter setting means 812 for assigning a parameter to each measurement point so as to make u or v of the parameter value uniform, and the addition direction and position of the node and the control point are automatically calculated by using the variance of the previous approximation error. Automatic node determining means 818 for determining, parametric curved surface approximating means 813 for sequentially approximating three-dimensional coordinate value data to a parametric curved surface using the aforementioned nodes, and data compressed by the parametric curved surface approximating means 813 is the data format of the input destination system. Data format conversion means 814 for converting the data into a data format and scanner control means 8 for controlling the entire system
By providing 15, the waviness that often occurs when approximating a curved surface is reduced, and the amount of data is efficiently reduced while the measured shape is saved, making it difficult to predict directly from the shape of the measured object in the past. The problem of insertion of the number of control points and the positions of nodes in order to obtain a more efficient approximation was solved, and three-dimensional CAD and computer graphics in a data format that is easy to change and correct the shape in the input destination system. Input can be made to the system 817.

(Third Embodiment) A third embodiment of the present invention will be described below with reference to the drawings.

FIG. 16 shows a third embodiment of the present invention.
It is a block connection diagram of a three-dimensional shape input device. In FIG. 16, reference numeral 1601 denotes a laser light source for generating slit light for irradiating the object to be measured, 1602 a mirror for changing the optical path of the slit light, 1603 for slit light for irradiating the object to be measured, 1604 for the object to be measured, 1605 to 1605. DUT 1604
X-axis moving mechanism for moving the X-axis at a constant pitch, 1
Reference numeral 606 is a camera for reading scattered light of the slit light 1603, 1607 is an A / D converter for converting the output signal of the camera 1606 into a digitized image signal, 1608 is an image memory for storing the image signal, and 1609 is a slit for the image signal. Slit scattered light center position detecting means for calculating the center position of scattered light, 1610 for coordinate calculation means for calculating three-dimensional coordinate values from the central position of slit scattered light, and 1612 for indicating a curve on the object to be measured. And the parameters of the parametric curved surface are u and v, the parameter setting means for assigning parameters to the respective measurement points so that the values u or v of the parameter values of the continuous measurement points on the curve are aligned, and the nodes and the control points And a node automatic determination means for automatically determining the direction and position to be added by using the variance of the previous approximation error. More parametric surfaces best approximation means for best approximates sequential parametric surface a 3-dimensional coordinate value data, a scanner control unit for controlling the whole system is 1616, 1617 the three-dimensional shape data obtained by the present apparatus,
1618 is a three-dimensional CAD system, and the above is the same as the configuration of FIG.

The difference from the configuration of FIG. 8 is that when three-dimensional coordinate value data is repeatedly approximated to a parametric curved surface until a curved surface with a specified accuracy or less is obtained, the first approximation is a uniform nodal point for the parametric curved surface approximation. After the second time, the three-dimensional coordinate value data is approximated to the parametric curved surface as it is, or the three-dimensional coordinate value data is mechanically irrespective of the shape of the DUT 1604 based on the approximation result up to the previous time. Three processing methods: dividing into two and then approximating to a parametric curved surface, or extracting a feature point sequence from a point group of three-dimensional coordinate value data and constructing a surface with a polygon patch using the feature points. The curved surface processing method determining means 1611 for determining which method to use in the above, and the three-dimensional coordinate value data representing the overall shape of the DUT 1604 are specified on the parametric curved surface. In the case of iteratively approximating a curved surface with accuracy or less, if the rate of decrease of the approximation error with the increase of the control points is low, one curved surface is divided and two different curved surfaces are approximated to the parametric curved surface. A region mechanical dividing unit 1613 that divides the three-dimensional coordinate value data into two in a direction having a large approximation error, and a measured three-dimensional coordinate value when the approximation error has a limit and the designated approximation accuracy cannot be achieved. A feature point patch creating means 1614 for extracting a feature point sequence from a point group of data and constructing a surface with a polygon patch using the feature points is further added. Further, in FIG. 16, the data format in FIG. 16 is added. The conversion means 814 is replaced by the parametric surface best approximation means 16
Surface and feature point patch creating means 1 constructed by 12
This is the point that the data format conversion means 1615 for converting the patch created by 614 into the data format of the input destination system is changed.

The operation of the three-dimensional shape input device configured as described above will be described. First, the DUT 80
4, the curved surface processing method determining means 1611 is a processing method for approximating the curved surface of the three-dimensional coordinate value data until a curved surface with a specified accuracy or less is obtained, and the first approximation is a uniform node. Parametric surface approximation is performed, and after the second time, based on the approximation results up to the previous time, the parametric surface best approximation means 1612 is performed, or the area mechanical dividing means 1613 is performed and then the parametric surface best approximation means 1612 is performed. Determines whether to perform the feature point patch creating means 1614. When it is determined by the curved surface processing method determining means 1611 to perform the parametric curved surface best approximating means 1612, the parametric curved surface best approximating means 1612 sets the parameters of the curved surface directing means and parametric curved surface for instructing a curve on the object to be measured as u and v. Then, the parameter setting means for giving a parameter to each measurement point so that the values u or v of the parameter values of the continuous measurement points on the curve are aligned, and the direction and the position at which the node and the control point are added are set to the previous approximation error. And the node automatic determination means for automatically determining using the variance of the three-dimensional coordinate value data are used to best approximate the three-dimensional coordinate value data to the parametric curved surface. Next, when it is determined that the curved surface processing method determining means 1611 performs the area mechanical dividing means 1613 and then the parametric curved surface best approximating means 1612, the area mechanical dividing means 1613 determines that the three-dimensional coordinate value data has a large approximation error. The area is divided into two, and the parametric curved surface best approximation means 1612 best approximates the divided areas to different curved surfaces. Next, the curved surface processing method determining means 1611 causes the feature point patch creating means 16
If it is determined to perform step 14, a feature point sequence is extracted from the point group of the measured three-dimensional coordinate value data, and the feature points are used to construct a surface with a polygon patch. The data format conversion means 1615 is the parametric curved surface best approximation means 161.
Curved surface and feature point patch creating means 16 constructed by 2
The patch created by 14 is converted into the data format of the input destination system, and the format-converted three-dimensional shape data 1617 is input to the three-dimensional CAD system 1618.

The detailed operation of the curved surface processing method determining means 1611 will be described below. In FIG. 17, in the curved surface processing method determining means 1611, the area mechanical dividing means 1 is used.
613 and the three-dimensional coordinate value data measured using the parametric surface best approximation means 1612 are Non-Uniform B-Spl which is one of the parametric surfaces.
ine (non-uniform bee spline) Surface approximation is performed with a precision not higher than the designated precision, or if the precision is not satisfied, the feature point patch creating means 1614
2 shows a flow for deciding whether to construct a patch using the feature points of the three-dimensional coordinate value data or which of the above processes should be performed. The parameters expressing the curved surface are u and v.

First, the three-dimensional coordinate values of the object measured in step 1701 are input. In step 1702, the number of control points in the u direction (n _u +1) and the rank m _u are designated. Step 1703
At, the value obtained by adding the number of control points and the rank m _u of the curved surface to the number of nodes is substituted. In step 1704, the node positions in the u direction are set at equal intervals. However, the nodes at both ends are multiplexed _mu times.
In step 1705, the number of control points in the v direction (n _v +1) and the rank m _v
Is specified. In step 1706, a value obtained by adding the number of control points and the rank m _v of the curved surface to the number of nodes is substituted. In step 1707, the v-direction node positions are set at equal intervals. However, the nodes at both ends are multiplexed m _v times. In Step 1708, Step 170
4. Using the nodes obtained in step 1707, the control points are calculated by approximating the data to (Equation 1) by least squares. In step 1709, the approximation error of the least square approximation performed in step 1708 is calculated. In step 1710, it is determined whether or not the approximation error is less than the specified tolerance. If the difference is less than the allowable error, the coordinate values of the control points and nodes in both the u and v directions calculated in step 1711 are output as parameters indicating the shape. If the tolerance is not satisfied in step 1710, the variance of the residual is calculated in both u and v directions in step 1712. When the allowable error is not satisfied in step 1713, 1 is added to the count Count. Step 171
If the count Count is equal to or smaller than the specified count Count _{0 in step} 4, it is assumed that the surface is not divided, and the calculation in step 1715 is performed. In step 1715, as the processing of the parametric surface best approximation means 1612, if the parameters of the curve designating means for designating the curve on the object to be measured and the parametric surface are u and v, u of the parameter values of the continuous measurement points on the curve are set. Alternatively, a parameter setting means for giving a parameter to each measurement point so as to make the values of v uniform, and a nodal point automatic decision means for automatically deciding the adding direction and position of the nodal point and the control point by using the variance of the previous approximation error. Perform processing. And
The procedure returns to step 1708, and the least squares approximation is performed again using the determined node positions. Next, when the allowable error is not satisfied in step 1714, the count Count is the specified count Cou.
When nt ₀ or more, the following calculation is performed. In step 1716, the process is branched into three stages depending on the magnitude of the reduction rate of the residual calculated in step 1712 in the Count count so far. If the rate of reduction of the residual is higher than the preset threshold value, the calculation of step 1715 is performed. If the rate of decrease is low, the process of the area mechanical dividing means 1612 is performed in step 1717. The area mechanical dividing means 1612 divides the three-dimensional coordinate value data into two in the direction in which the residual difference is large. And
Hereinafter, iterative curved surface approximation is newly performed with each surface as a different surface. However, the number of control points in the divided direction is the half of the number of control points approximated the previous time plus the rank, and the same number is used in the direction not divided. In step 1718, Count is reset to the initial value 0. If the rate of decrease is considerably low and the surface is so small that it cannot be further divided, the processing of the feature point patch creating means 1614 is performed in step 1719. Feature point patch creating means 1614
Generates a polygonal patch using the feature points extracted from the data points. By repeating the above procedure, an approximate curved surface that satisfies the specified tolerance can be realized with the least amount of data.

The operation of the feature point patch creating means 1614 in step 1719 of FIG. 17 will be described in detail. First, the three-dimensional coordinate value of a point sequence in a certain divided plane is input. Next, the feature points are extracted in each slit of the input three-dimensional coordinate values. Then, using the extracted feature points for each slit, a quadrangular patch or a triangular patch is created between the adjacent slits.

Next, the details of the operation of extracting the characteristic points in the characteristic point patch creating means 1614 will be described. FIG. 18 shows a detailed flow of feature point extraction.

First, in step 1801, the three-dimensional coordinate value of a point on one slit is input. In step 1802,
The equation of a straight line connecting both ends of the point sequence input in step 1801 is calculated. In step 1803, the distance between each measurement point excluding both ends and the straight line calculated in step 1801 is calculated. In step 1804, the maximum distance among the distances between the points and the straight line obtained in step 1803 is calculated, and the point that gives the maximum distance is set as the characteristic point. In step 1805, the characteristic point and both end points are connected by a polygonal line, and the equation of each line segment is obtained. FIG. 19 shows the procedure 1805.
Represents the operation of. In Step 1806, Step 18
The residual difference between the sampling point and the measurement point on each line segment calculated in 05 is calculated. In step 1807, if the residual is less than or equal to the threshold value representing the preset approximation accuracy, the feature point extraction ends. If it is equal to or more than the threshold value, in step 1808, the distance between the line segment calculated in step 1805 and the measurement point excluding both end points is calculated, and the procedure returns to step 1804 to set the measurement point that brings the maximum distance as a new feature point. To do. By repeating the above operation, the feature points can be extracted from the point sequence.

Next, the operation of the feature point patch creating means 1614 to create a polygonal patch will be described in detail. Figure 2
0 indicates the flow of polygon patch generation between two adjacent slits.

First, in step 2001, three-dimensional coordinate values of feature points in two adjacent slits are input. Step 20
At 02, two quadrangle points are created in order from the top at the feature points on the slit to create a temporary quadrangle. In step 2003, the lengths of the two diagonal lines of the quadrangle are calculated. In step 2004, it is determined whether or not the difference between the lengths of the two diagonal lines is greater than or equal to the threshold value. If it is within the threshold value, in step 2005, a square patch is created using the four points. If it is equal to or larger than the threshold value, in step 2006, a triangular patch is created with three points excluding the lower one point forming the longer diagonal line. FIG. 21 shows an example in which a triangular patch is created. 2101, 2
102, 2103, and 2104 are 4 in step 2002.
The vertices 2105 and 2106 of the polygon are points 2101 and 21.
02, a point 2103, and a point 2104 are diagonal lines of a quadrangle. Since there is a large difference in the lengths of the diagonal line 2105 and the diagonal line 2106, the three points except the point 2104 are 3
A square patch is created. Next, in step 2007, it is determined whether or not two points for forming the quadrangle in step 2002 remain on one of the slits. If there are two points, the procedure returns to step 2002, and if there is only one point, step 200
At 8, all the remaining patches form a triangular patch with one point as the apex.

As described above, according to this embodiment, the laser light source 1601 and the mirror 1602, which are slit light sources for generating the laser slit light for irradiating the object to be measured, and the object 1604 to be measured are arranged at a constant pitch in the X-axis direction. An X-axis moving mechanism 1605 that moves, a camera 1606 that captures the scattered light of the slit light by the DUT 1604, and a camera 1606.
A to convert the output signal from the device into a digitized image signal
/ D converter 1607, an image memory 1608 for storing the image signal, slit scattered light center position detection means 1609 for detecting the center position of the slit scattered light from the image signal stored in the image memory 1608, slit scattered light When the coordinate calculation means 1610 for calculating the three-dimensional coordinate value of the object 1604 from the center position and the three-dimensional coordinate value data are repeatedly approximated to the parametric curved surface until a curved surface with a specified accuracy or less is obtained, the first time. For the approximation, the parametric surface approximation is performed using uniform nodes, and after the second time, the 3D coordinate value data is approximated to the parametric curved surface as it is based on the approximation results up to the previous time, or the 3D coordinate value data is measured. Whether it is mechanically divided into two parts regardless of the shape of 1604 and then approximated to a parametric surface,
Alternatively, a curved surface processing method for deciding which method is used among three processing methods of extracting a characteristic point sequence from a point group of three-dimensional coordinate value data and constructing a surface with a polygon patch using the characteristic points When u and v are parameters of the determining means 1611 and the curve designating means for designating a curve on the object to be measured and the parametric curved surface, the values u or v of the parameter values of the continuous measuring points on the curve are aligned. The node is used by providing a parameter setting means for assigning a parameter to each measurement point and a node automatic determination means for automatically determining the direction and position of addition of the node and the control point using the variance of the previous approximation error. Parametric surface best approximation means 1612 for sequentially approximating the three-dimensional coordinate value data to a parametric curved surface sequentially, and three-dimensional coordinate value data representing the entire shape of the DUT 1604. When iteratively approximating to a parametric surface until a surface of less than the specified accuracy is obtained, if the rate of decrease of the approximation error due to the increase of control points is low, one surface is divided into two different parametric surfaces. Region mechanical dividing means 1 which divides the three-dimensional coordinate value data into two in the direction of a large approximation error in order to approximate
613, and when the approximation error has a limit and the designated approximation accuracy cannot be achieved, a feature point sequence is extracted from the point group of the measured three-dimensional coordinate value data, and the feature points are used to form a polygon patch. Feature point patch creating means 1614 for constructing a surface and parametric curved surface best approximation means 16
Surface and feature point patch creating means 1 constructed by 12
Data format conversion means 1615 for converting the patch created by 614 into the data format of the input destination system, and scanner control means 161 for controlling the entire system.
6 and three-dimensional shape data 1617 obtained by this device,
By providing the three-dimensional CAD system 1618, it is possible to reduce the waviness shape that often occurs at the time of curved surface approximation, efficiently reduce the amount of data while maintaining the measured shape, and obtain a curved surface with a specified accuracy or less. In the past, it was difficult to predict directly from the shape of the measured object, but it solved the problem of inserting the number of control points and the positions of the nodes to obtain a more efficient approximation, and changed the shape in the input system.
The data can be input to the three-dimensional CAD or computer graphics system 1618 in a data format that can be easily modified.

(Embodiment 4) A fourth embodiment of the present invention will be described below with reference to the drawings.

FIG. 22 shows a third embodiment of the present invention.
It is a block diagram of a three-dimensional shape input device. In FIG. 22, 2
Reference numeral 201 is a laser light source for generating slit light for irradiating the DUT 2204, 2202 is a mirror for changing the optical path of the slit light, 2203 is slit light for irradiating the DUT 2204, 2205 is the X-axis direction of the DUT 2204. 2206 is a camera for reading scattered light of the slit light 2203, 2207 is an A / D converter for converting an output signal of the camera 2206 into a digital image signal, and 2208 is an image signal. An image memory for storing 2209 is a slit scattered light center position detecting means for calculating the central position of slit scattered light from an image signal, 2210 is a coordinate calculating means for calculating a three-dimensional coordinate value from the central position of the slit scattered light, and 2211 is a target. Curve indicating means for indicating a curve on the object to be measured 2212 is a parameter of the parametric surface u and v are parameter setting means for assigning a parameter to each measurement point so that the values u or v of the parameter values of the measurement points on the continuous curve are aligned, and 2215 is a scanner control means for controlling the entire system. Two
217 is three-dimensional shape data obtained by this device, 2218
Is a three-dimensional CAD system, and the above is the same as the configuration of FIG.

1 is different from the parametric curve approximating means 113 in FIG. 1 in that the three-dimensional coordinate value data representing the entire shape of the object to be measured is approximated to a parametric curve for each slit. 221
3 is added, and interpolation surface generation means 2214 for creating a curved surface for interpolating the parametric curve of each slit is added,
This is the point that the data format conversion means 114 is changed to the data format conversion means 2216 for converting the data compressed by the interpolation curved surface generation means 2214 into the data format of the input destination system.

Instead of the X-axis moving mechanism 2205 which moves the object to be measured 2204, slit light is emitted from the object to be measured 2
A slit light scanning means for scanning on 204 may be used.

The operation of the three-dimensional shape input device configured as described above will be described. First, the DUT 22
The process is the same as that of the first embodiment until the entire 04 three-dimensional shape data is acquired.

Next, the parametric curve approximation means 221
3 approximates the three-dimensional coordinate value to a parametric curve for each slit. The interpolation surface generation unit 2214 creates a surface for interpolating the parametric curve of each slit obtained by the parametric curve approximation unit 2213 by the following procedure. First, connect the vertices of adjacent spline curves.
The edges to be generated are smooth curves that pass through the corresponding vertices of adjacent curves. An interpolated curved surface is generated by interpolating a curved surface on the ridge thus generated. The data format conversion means 2216 is the interpolation surface generation means 2
The data compressed by 214 is converted into the data format of the input destination system, and the format-converted three-dimensional shape data 2217 is converted into the three-dimensional CAD system 2218.
Entered in.

Hereinafter, the parametric curve approximation means 221
A more detailed operation in 3 will be described. In FIG. 23, data of one slit in the parametric curve approximating means 2213 is Non- which is one of parametric curves.
The flow of the method of approximating below the approximation precision designated to the Uniform B-Spline (non-uniform bee spline) curve is shown.

First, the YZ coordinate value of one slit is input from the three-dimensional coordinate value data measured in step 2301. In step 2302, the secondary difference value is calculated. Step 23
In 03, the number of control points (n + 1) and the rank m are designated. Step 2
At 304, the maximum point of the secondary difference value is set as a node. Step 2
At 305, a value obtained by adding the number of control points and the rank m of the curve to the number of nodes is substituted. In step 2306, the control points are calculated by using the nodes obtained in step 2305 and approximating the data to (Equation 4) by least squares. In (Equation 4), P _i is a control point, N _{i, p} is a spline basis function, and t is a parameter.

[0063]

[Equation 4]

In step 2307, the approximation error of the least square approximation performed in step 2306 is calculated. In step 2308, it is determined whether or not the approximation error is less than the specified allowable error. If the approximation error is less than the specified tolerance, the coordinate value of the node and the coordinate value of the control point obtained in step 2309 are output as parameters indicating the shape of the slit. On the other hand, if the allowable error is not satisfied in step 2308, 1 is added to the number of node movements in step 2311, and
At 12, the point with the largest error is calculated from the data points and the approximate curve, and the sum of the left and right errors centered on that node is compared,
Move the node a minute distance in the direction with a large error. on the other hand,
If the allowable error is not satisfied by the designated number of movements, the number of control points is incremented by 1 in step 2313. A new node is added at the position with the largest error. And step 230
Return to 5.

By the above operation, an approximate curve satisfying the specified tolerance can be realized with the minimum data amount. However, regarding the calculation of the nodes in step 2304, there is a method in which the quadratic difference value is approximated to a broken line and the broken points are used as the nodes. Further, it is possible to obtain a more accurate maximum point by approximating the data with a multiple-order function in advance and taking the second derivative of the function instead of using the difference value. It does not limit the invention. Further, although the curve approximation is performed for each slit, depending on the shape of the object to be measured, it may be more appropriate to perform curve approximation by connecting the nth point of each slit to generate one line of data. is there. It goes without saying that data can be approximated by the same procedure in this case as well. Further, when the approximation error is reduced by repeating the movement of the nodes and the increase of the number of control points, if the rate of decrease of the approximation error is low even if the control points are increased, the curve is divided into a plurality of curves, each of which is different. By approximating the parametric curve as a curve, a curve having a specified approximation error or less can be obtained. Also, by approximating each curve in parallel, the time required for approximation can be significantly reduced.

As described above, according to this embodiment, the laser light source 2201 and the mirror 2202, which serve as a slit light source for generating the laser slit light for irradiating the object to be measured, and the object 2204 to be measured at a constant pitch in the X-axis direction. An X-axis moving mechanism for moving, a camera 2206 for imaging scattered light of slit light by the DUT 2204, an A / D converter 2207 for converting an output signal from the camera 2206 into a digitized image signal, and the above image. Image memory 22 for storing signals
08, the slit scattered light center position detecting means 2209 for detecting the center position of the slit scattered light from the image signal stored in the image memory 2208, and the three-dimensional coordinate value of the DUT 2204 is calculated from the slit scattered light center position. Assuming that the coordinate calculating means 2210, the curve indicating means 2211 for indicating a curve on the object to be measured, and the parameters of the parametric curved surface are u and v, u or v of the parameter values of the continuous measurement points on the curve. Parameter setting means 2212 for assigning parameters to each measurement point so as to align the same, parametric curve approximating means 2213 for approximating three-dimensional coordinate value data representing the overall shape of the measured object to a parametric curve for each slit, and each slit. Interpolation surface generation means 2214 for creating a curved surface for interpolating the parametric curve of
By providing the scanner control means 2215 for controlling the entire system and the data format conversion means 2216 for converting the data compressed by the interpolated curved surface generation means 2214 into the data format of the input destination system, the waviness shape that often occurs at the time of curved surface approximation can be obtained. The amount of data is efficiently reduced while saving the reduced and measured shapes, the calculation is easy because each curve is approximated independently, and the parallel processing can reduce the calculation time, and the input destination Three-dimensional CAD and computer graphics system 2 in a data format that can easily change and modify the shape in the system
218 can be entered.

(Fifth Embodiment) A fifth embodiment of the present invention will be described below with reference to the drawings.

FIG. 24 shows a third embodiment of the present invention.
It is a block connection diagram of a three-dimensional shape input device. In FIG. 24, 2401 is a laser light source for generating slit light for irradiating the object to be measured, 2402 is a mirror for changing the optical path of the slit light, 2403 is slit light for irradiating the object to be measured, 2404 is the object to be measured, and 2405 is DUT 2404
X-axis moving mechanism for moving the X-axis at a constant pitch, 2
406 is a camera for reading scattered light of the slit light 2403, 2407 is an A / D converter for converting the output signal of the camera 2406 into a digitized image signal, 2408 is an image memory for storing the image signal, and 2409 is a slit for the image signal. Slit scattered light center position detecting means for calculating the central position of scattered light, 2410 is coordinate calculating means for calculating three-dimensional coordinate values from the central position of slit scattered light, 2416 is a scanner controlling means for controlling the entire system, and 2418 is a book. 3D shape data obtained by the device, 2419 is 3D CAD
In the system, the above is the same as the configuration of FIG.

8 is different from the configuration of FIG. 8 in that the reflectance of the slit light with respect to the object 804 of FIG. 8 is different from the surface color of the object of measurement, and two colors having different reflectances are used. A curve for dividing an area for dividing the object to be measured 2404 into a plurality of areas by each color and a curve for setting parameters when approximating each area to a curved surface are measured by a person. The object 2404 has a different reflectance on the object to be measured with respect to the curve indicating means 811 of FIG.
A point changed to a curve designating means A2411 for designating two types of curves written in one color, and a curve designating means A2411.
With respect to the point added with the area dividing means 2412 for dividing the three-dimensional coordinate value data according to the curve for area division designated by the above, and the parameter setting means 812 of FIG. If the parameters of the curved surface are u and v, the curve designating means A241
Parameter setting means 24 for giving a parameter to each measurement point so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by 1 are made uniform.
In FIG. 13, the node automatic determination means 818 of FIG. 8 automatically determines the direction and position of adding the node and the control point in each of the divided areas by using the variance of the previous approximation error. The deciding means 2414, in contrast to the parametric curved surface approximating means 813 of FIG. 8, uses the nodes in each of the divided areas to sequentially approximate the three-dimensional coordinate value data of each area to the parametric curved surface approximating means 2415. 8 is different from the data format conversion means 814 of FIG. 8 in that it is changed to data format conversion means 2417 for converting the data compressed by the parametric curved surface approximating means 2415 into the data format of the input destination system. is there.

The operation of the three-dimensional shape input device configured as described above will be described. First, the measured object 24
If 04 is brown, the person using the black, which has a lower reflectance for the slit light than brown, measures the object to be measured 24.
A curve for dividing an area is written on 04, and a curve for setting parameters on the DUT 2404 is written using white, which is a color having a higher reflectance than brown for slit light. The area dividing line is drawn in a portion where it is considered that the correction of the connection between the planes will be applied after the CAD input. X
An object to be measured 2404 is installed on the axis moving mechanism 2405.
The slit light 2403 from the laser light source 2401 has its optical path changed by the mirror 2402, and is irradiated onto the DUT 2404 on the X-axis moving mechanism 2405. Below, the DUT 240
Example 1 until the entire three-dimensional shape data of No. 4 is acquired
It is the same operation as. The curve designating means A 2411 has a curve for dividing an area drawn on the DUT 2404 and a curve for setting a parameter for each color based on the three-dimensional coordinate value data representing the entire shape of the measured DUT. Recognized by the difference in reflection brightness, the area dividing means 2412 uses the curve extracted by the curve indicating means A 2411 to determine 3
The dimensional coordinate value data is divided into a plurality of areas. The parameter setting means 2413, the automatic node determination means 2414, the parametric curved surface approximating means 2415 are the area dividing means 2412.
The shape information is compressed by approximating each area extracted by the method to a parametric curved surface individually as in the second embodiment. The data format converting means 2417 converts the parameter values of the plurality of parametric curved surfaces representing the overall shape of the DUT 2404 approximated by the parametric curved surface approximating means 2415 into a data format for the CAD system of the input destination, and the format is converted. The three-dimensional shape data 2418 is input to the three-dimensional CAD system 2419.

A more detailed operation of the curve designating means A 2411 will be described. First, the maximum luminance value is obtained for each horizontal scanning line of the camera in the image temporarily stored in the image memory 2408, and this is set as the luminance value at the measurement point. By repeating this for each image, the luminance value at each measurement point is obtained. DUT 2404
Since the curves for area division and parameter setting described above use colors having different reflectances for slit light, threshold values d1 and d2 are set in advance, and the brightness value d at the measurement point is The measurement points with d> d1 belong to the set D1, the measurement points with d1 ≧ d> d2 belong to the set D2,
The measurement points with d2 ≧ d belong to the set D3. Then, by measuring the measurement points belonging to the set D1, the DUT 240
4. The curve for parameter setting specified above can be extracted. The extracted curve is subjected to expansion processing and thinning processing to create a continuous curve with one width. In addition, by measuring the measurement points belonging to the set D3, the measured object 24
It is possible to extract a curve for area division indicated by 04.
In the method of instructing a curve for parameter setting in the curve instructing means A, a rough curve is specified on the object to be measured using a color whose reflectance to slit light is different from the surface color of the object to be measured. The curve is extracted from a portion of the image signal having a different luminance value and a peripheral portion thereof based on the edge-likeness obtained by using the three-dimensional coordinate value data, or the obtained curve is an edge portion of a measured object. The edge of the object to be measured using the three-dimensional coordinate value data without any manual intervention, or the camera reads the slit scattered light and does not project the slit light. Outputs depending on whether the camera captures the slit light or the entire brightness information of the DUT that does not project the slit light. Memory switching means for switching, a first image memory for storing an image signal of slit scattered light by switching of the memory switching means, and the entire object to be measured not projecting slit light by switching of the memory switching means A second image memory for storing the luminance image is separately provided, and the obtained curve is regarded as the edge portion of the object to be measured, and whether the edge is automatically obtained using the luminance image without human intervention. , Or a color that specifies the curve on the DUT using a color different from the surface color of the DUT, and the camera reads the slit scattered light and images the entire DUT without projecting the slit light. Output switching depending on whether the camera captures the slit light or the entire brightness information of the DUT that does not project the slit light. Memory switching means, a first image memory for storing an image signal of slit scattered light by switching of the memory switching means, and brightness of the entire object to be measured not projecting slit light by switching of the memory switching means A second image memory for storing an image is provided separately, and the curve is extracted from a luminance image using a color difference, or measured three-dimensional coordinate value data is displayed on a computer, and the displayed three A human may use the mouse to indicate a curve for the dimensional coordinate value data.

Next, the parametric curved surface approximating means 241.
A more detailed operation in 5 will be described. First, the boundary of the region is traced, and all points where three or more curved surfaces intersect are extracted. The boundary of the area is divided by the obtained points, and the point sequence on the divided boundary is automatically determined by the node automatic determination means 24.
The nodal point determined in 14 is used to approximate the spline curve. When the approximation accuracy is low, the boundary is further divided into two and each is approximated to a spline curve. The approximation method is similar to that of the parametric curve approximation means 2213. Next, the control point obtained by the curve approximation is known, and the point sequence in the area is approximated to the spline curved surface. The curved surface approximation is similar to the method described in the second embodiment. At this time, the known control points are the end points of the control point sequence on the curved surface. When the region boundary is divided into three curves, one of the end points of the curve is overlapped with an arbitrary control point, and least-squares approximation is performed. Further, in the case of being divided into five or more, first, a perpendicular bisector of a straight line that connects both end points at the boundary approximated to any two spline curves is calculated, and the intersection point is obtained. Next, 2 next to each other
A common end point of each curve, the middle point of each curve, and the above-mentioned four points of intersections of the calculated vertical bisectors are set as vertices to form one small area, and each small area is approximated to a spline curved surface. Therefore, adjacent small areas have a common boundary curve, so that the boundaries of the curved surfaces are the same control points. For example, if it is a pentagon,
It is composed of five small area spline curved surfaces. In addition,
As for the approximation of each region, as shown in the third embodiment, the first approximation is a parametric curved surface approximation with uniform nodes, and the second and subsequent approximations are based on the approximation results up to the previous time, and three-dimensional coordinate value data is obtained. It is approximated to the parametric curved surface as it is, or the three-dimensional coordinate value data is measured 24
No. 04 mechanically divides into two and then approximates to a parametric curved surface, or extracts a feature point sequence from a point group of three-dimensional coordinate value data and uses the feature points to form a surface with a polygon patch. The three-dimensional coordinate value data may be repeatedly approximated to a parametric curved surface until a curved surface with a specified accuracy or less is obtained by deciding which method to use in the three processing methods of construction.

As described above, according to this embodiment, two colors having different reflectances for slit light from the surface color of the object to be measured and having different reflectances to each other are used, and the object to be measured is respectively colored. An object to be measured 240 in which a person enters a curve for dividing an area for performing approximation by dividing into a plurality of areas and a curve for setting parameters for approximating each area to a curved surface.
4, a laser light source 2401 and a mirror 2402 as a slit light source for generating laser slit light for irradiating the DUT 2404, an X-axis moving mechanism 2405 for moving the DUT 2404 at a constant pitch in the X-axis direction, A camera 2 for capturing the scattered light of the slit light by the measurement object 2404.
406, an A / D converter 2407 that converts an output signal from the camera 2406 into a digitized image signal, an image memory 2408 that stores the image signal, and slit scattered light from the image signal stored in the image memory 2408. Slit scattered light center position detecting means 24 for detecting the center position of
09, and the measured object 2404 from the slit scattered light center position
The coordinate calculating means 2410 for calculating the three-dimensional coordinate values of 2 and the two filled in two colors having different reflectances on the measured object.
In each of the divided areas, a curve instruction means A2411 for instructing a kind of curve, an area dividing means 2412 for area-dividing the three-dimensional coordinate value data according to the curve for area division instructed by the curve instruction means A2411. If the parameters of the parametric curved surface are u and v, the parameters are given to the respective measurement points so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by the curve designating means A2411 are made uniform. Parameter setting means 2413, and automatic nodal point determining means 2414 for automatically determining the direction and position of addition of the nodal point and control point in each of the divided areas using the variance of the previous approximation error. The 3D coordinate data of each area is successively approximated to a parametric surface using the nodes in each area. Parametric surface approximation means 241
5, a data format conversion unit 2417 for converting the data compressed by the parametric curved surface approximation unit 2415 into the data format of the input destination system, and a scanner control unit 2416 for controlling the entire system. Input data to a three-dimensional CAD or computer graphics system 2419 in a data format that can efficiently reduce the amount of data while saving the measured shape, and can easily change or correct the intended shape in the input destination system. be able to.

(Embodiment 6) A sixth embodiment of the present invention will be described below with reference to the drawings.

FIG. 25 shows a third embodiment of the present invention.
It is a block diagram of a three-dimensional shape input device. In FIG. 25, 2
Reference numeral 501 denotes a laser light source for generating slit light for irradiating the DUT 2504, 2502 for a mirror for changing the optical path of the slit light, 2503 for slit light for irradiating the DUT 2504, 2505 for the X-axis direction of the DUT 2504. 2506 is a camera for reading the scattered light of the slit light 2503, 2507 is an A / D converter for converting the output signal of the camera 2506 to a digitized image signal, and 2508 is the image signal. An image memory for storing, 2509 a slit scattered light center position detecting means for calculating the central position of the slit scattered light from the image signal, 2510 a coordinate calculating means for calculating a three-dimensional coordinate value from the central position of the slit scattered light, and 2516 for the whole. Scanner control means for controlling the system, 2518 is provided by this device 3
Dimensional shape data, 2519 is a three-dimensional CAD system,
The above is the same as the configuration of FIG. In this embodiment, the laser light source 2501 is He-Ne having a wavelength of 633 nm.
A (helium-neon) laser was used. The wavelength of the laser used as the laser light source 2501 does not limit the present invention.

The difference from the configuration of FIG. 24 is that, for the DUT 2404 in FIG. 24, two colors having different reflectances for slit light from the surface color of the DUT and further having different reflectances are used. A person roughly fills out a curve that divides the area for dividing the measured object into multiple areas with each color and a curve for setting the parameters when approximating each area to the curved surface. DUT 250
The point changed to 4, the curve indicating means A241 in FIG.
On the other hand, the point is changed to the curve designating means B2511 for extracting the area dividing line based on the edge-likeness obtained from the measurement data from the pixel whose luminance value of the image signal is different from the surface color of the DUT 2504 and its peripheral pixels. 2 is different from the area dividing means 2412 in FIG. 24 in that the area dividing means 2512 divides the three-dimensional coordinate value data into areas according to the curve for area division indicated by the curve indicating means B.
4, the parameter u of the parametric curved surface in each of the divided areas is set to u and v to the parameter setting means 2413 in FIG. Alternatively, the point is changed to the parameter setting means 2513 for giving a parameter to each measurement point so as to make the values of v uniform, and the node automatic control means 2414 in FIG. 24 and the parametric curved surface approximating means 24 in FIG. 24 in which the automatic addition means 2514 for automatically determining the direction and position of addition of points using the variance of the previous approximation error.
On the other hand, the parametric curved surface approximating means 251 for sequentially approximating the three-dimensional coordinate value data of each area to the parametric curved surface using the nodes in each of the divided areas.
5 and the data format conversion means 24 in FIG.
17 is that the divided areas are changed to data format conversion means 2517 for converting the data compressed by the parametric curved surface approximation means 2515 into the data format of the input destination system.

The operation of the three-dimensional shape input device configured as described above will be described. First, the measured object 24
Assuming that 04 is brown, a person uses the black, which has a lower reflectance for the slit light than brown, to be measured 12
A rough curve for dividing the area on 04 is written thickly, and a rough curve for setting parameters on the DUT 1204 using white, which is a color having a higher reflectance than brown for slit light. Fill in a thick curve. The curve for the area division is drawn in a portion where it is likely that the correction for the connection between the planes will be applied after CAD input. It is a part mainly corresponding to an edge. The object to be measured 2504 is installed on the X-axis moving mechanism 2505. The slit light 2503 from the laser light source 2501 has its optical path changed by the mirror 2502 and is irradiated onto the object to be measured 2504 on the X-axis moving mechanism 2505. Hereinafter, the same operation as that of the first embodiment is performed until the entire three-dimensional shape data of the DUT 2504 is acquired. The curve designating means B2511 divides the area drawn on the object to be measured 2504 from the three-dimensional coordinate value data representing the measured overall shape of the object to be measured, and a curve for setting a parameter for each color. Recognized by the difference in reflected brightness,
Further, a curve for area division and a curve for parameter setting are extracted from pixels having different reflection brightness and peripheral pixels thereof based on the edge likelihood obtained from the measurement data. Hereinafter, in the same manner as the fifth embodiment, the area dividing unit 251.
Reference numeral 2 divides the three-dimensional coordinate value data into a plurality of areas by the curve for area division extracted by the curve designating means B2511, and the parameter setting means 2513, automatic node determination means 2514, and parametric curved surface approximating means 2515 are area dividing means. The shape information is compressed by individually approximating each area extracted by 2512 to a parametric curved surface in the same manner as in the second embodiment, and the data format conversion means 2517 is the whole measured object 2504 approximated by the parametric curved surface approximating means 2515. CAD of the input destination of the parameter values of a plurality of parametric surfaces representing the shape of
The three-dimensional shape data 2518 converted into the data format for the system and the format is converted into the three-dimensional CAD
Input to the system 2519.

The detailed operation of the curve designating means B2511 will be described. First, the maximum luminance value is obtained for each horizontal scanning line of the camera in the image temporarily stored in the image memory 2508, and this is set as the luminance value at the measurement point. By repeating this for each image, the luminance value at each measurement point is obtained. DUT 2504
Since the curves for area division and parameter setting described above use colors having different reflectances for slit light, threshold values d1 and d2 are set in advance, and the brightness value d at the measurement point is The measurement points with d> d1 belong to the set D1, the measurement points with d1 ≧ d> d2 belong to the set D2,
The measurement points with d2 ≧ d belong to the set D3. Then, the sequence of measurement points belonging to the set D1 forms a region on the DUT 2504 that includes a curve for instructing parameter setting. This is because the curve specified by humans is roughly and thickly written. Similarly, the measurement points belonging to the set D3 form an area on the object to be measured 2504 that includes the indicated curve for area division. Then, a curve for dividing the area and a curve for parameter setting are extracted based on the edge-likeness obtained from the measurement data in each area. Figure 2
6 shows a flow of a method of extracting a curve for dividing a region from the set D3 based on the edge-likeness obtained from the measurement data from two types of pixels having brightness values different from those of the surface of the object to be measured and peripheral pixels. A curve for parameter setting can be similarly extracted from the set D1.

First, in step 2601, the image memory 2508
The image signal stored in is input. In step 2602, the pixels included in D3 are extracted. In step 2603, the 8-neighbor expansion process is performed on the pixels extracted in step 2602. Since the obtained image is an expansion of the area dividing instruction line, it represents the area including the true area dividing line. This is called the area surrounding the area dividing line. In step 2604, the depth z of the three-dimensional coordinate value data of the point sequence in the area around the area dividing line
An edge detection operator is applied to the coordinate values. The value obtained as a result is called an edge strength representing edge-likeness. A Kirsch operator is shown in FIG. 27 as an edge detection operator. In step 2605, a pixel having the weakest edge strength is searched for among all the pixels in the area surrounding the area dividing line, and the pixel of interest is determined, and it is determined whether or not the pixel of interest can be deleted. The erasability is such that there are only two pixels within 8 neighborhoods of the pixel of interest, and if the two pixels are not adjacent, the pixel of interest is considered to be erasable, and in other cases, it is deleted. An example of erasability is shown in FIG. Step 2605
Is repeated until there are no more pixels that can be deleted (procedure 260).
By 6), the area around the area dividing line can be thinned. The thinned result by the above operation is used as a curve for area division or a curve for parameter setting.

In the curve designating means B2511,
The method of designating a curve for parameter setting is such that the reflectance for slit light is different from the surface color of the DUT, the curve is specified on the DUT, and the brightness value is different in the image signal. A curve indicating means A2411 for extracting a portion as the curve, or the curved line to be obtained is regarded as an edge portion of the object to be measured, and an edge is automatically obtained by using the three-dimensional coordinate value data without manual intervention. Alternatively, the camera reads the slit scattered light and captures the entire DUT without projecting the slit light, and the camera captures the slit light or does not project the slit light. A memory switching unit that switches the output depending on whether the brightness information of the entire measurement object is imaged, and an image signal of the slit scattered light by switching the memory switching unit A first image memory to be memorized and a second image memory for storing a luminance image of the entire object to be measured which is not projecting slit light by switching the memory switching means are separately provided, and the obtained curve is It is regarded as the edge part of the object to be measured, and the edge is automatically obtained using the luminance image without human intervention, or the curve is measured on the object to be measured using a color different from the surface color of the object to be measured. Is a color camera that reads the slit scattered light by the camera and also captures the entire DUT in a state where the slit light is not projected, and the camera either captures the slit light or does not project the slit light. A memory switching unit that switches the output depending on whether or not the brightness information of the entire object to be measured is imaged, and the image signal of the slit scattered light is stored by switching the memory switching unit. 1 image memory and a second image memory for storing a brightness image of the entire object to be measured which is not projecting slit light by switching the memory switching means are separately provided, and the difference in color from the brightness image is used. The curve extracted or measured 3
The dimensional coordinate value data may be displayed on a computer, and a human may use the mouse to indicate a curve for the displayed three-dimensional coordinate value data.

As described above, according to the present embodiment, two colors having different reflectances for slit light from the surface color of the object to be measured and having different reflectances from each other are used, and the object to be measured is respectively colored. An object to be measured 2504 in which a person roughly and roughly writes a curve for dividing an area for performing approximation by dividing into a plurality of areas and a curve for setting parameters when approximating each area to a curved surface, A laser light source 2501 as a slit light source for generating a laser slit light for irradiating a measurement object 2504 and a mirror 2502, and a measurement object 2504.
X-axis moving mechanism 25 for moving the X-axis in a constant pitch
05, a camera 2506 that captures the scattered light of the slit light by the DUT 2504, an A / D converter 2507 that converts the output signal from the camera 2506 into a digitized image signal, and an image that stores the image signal. Memory 250
8, a slit scattered light center position detecting means 2509 for detecting the center position of the slit scattered light from the image signal stored in the image memory 2508, and a three-dimensional coordinate value of the object to be measured 2504 is calculated from the slit scattered light center position. A curve for area division and curved surface approximation for each area based on the edge likelihood calculated from the measurement data from the coordinate calculation means 2510, the pixel whose luminance value of the image signal is different from the surface color of the DUT 2504, and its peripheral pixels. Curve designating means B2511 for extracting a curve used for parameter setting at the time, and curve designating means B2
Area dividing means 25 for dividing the three-dimensional coordinate value data into areas according to the area dividing curve indicated by 511.
12 and the parameters of the parametric curved surface in each of the divided areas are u and v, the curve designating means B2.
In each of the divided areas, a parameter setting means 2513 for giving a parameter to each measurement point so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by 511 are made uniform. Node and control point addition direction and position are automatically determined using the previous approximation error variance, and node automatic determination means 2514 is provided.
Parametric surface approximation means 2515 for sequentially approximating the three-dimensional coordinate value data of each area into a parametric surface using the nodes in each of the divided areas, and parametric surface approximation means 25 for each of the divided areas.
Data format conversion means 2 for converting the data compressed by 15 into the data format of the input destination system
517 and scanner control means 2516 for controlling the entire system
By providing the 3D CAD and computer graphics system in the data format, the amount of data can be efficiently reduced while the measured shape is saved and the intended shape can be easily changed or modified in the input destination system. Two
519 can be entered.

(Embodiment 7) Hereinafter, a seventh embodiment of the present invention will be described with reference to the drawings.

FIG. 29 shows a third embodiment of the present invention.
It is a block diagram of a three-dimensional shape input device. In FIG. 29, 2
Reference numeral 901 denotes a laser light source for generating slit light for irradiating the DUT 2904, 2902 for a mirror for changing the optical path of the slit light, 2903 for slit light for irradiating the DUT 2904, 2905 for the X axis direction of the DUT 2904. 2906 is a camera for reading scattered light of the slit light 2903, 2907 is an A / D converter for converting the output signal of the camera 2906 into a digitized image signal, and 2908 is an image signal An image memory for storing, 2909 a slit scattered light center position detecting means for calculating a central position of slit scattered light from an image signal, 2910 a coordinate calculating means for calculating a three-dimensional coordinate value from the central position of the slit scattered light, and 2916 an overall The scanner control means for controlling the system, 2918, is provided by this device.
Dimensional shape data, 2919 is a three-dimensional CAD system,
The above is the same as the configuration of FIG.

The difference from the configuration of FIG. 24 is that the object to be measured 2404 in FIG. 24 is changed to an object to be measured 2904 in which a curve is not entered by a person, and the curve indicating means A 2411 in FIG. 24 is measured. A curve that divides the shape of the measured object into regions is automatically extracted using the three-dimensional coordinate value data representing the entire shape of the measured object, and the threshold value is relaxed for each region and the same operation is performed. The point changed to the curve designating means C2911 for extracting the curve for setting the parameter for approximating each area to the curved surface by performing the operation shown in FIG.
The area dividing means 2412 in FIG. 4 is changed to an area dividing means 2912 for dividing the three-dimensional coordinate value data into areas according to the curve for area division indicated by the curve indicating means C2911, and the parameter setting means 2413 in FIG. On the other hand, if the parameters of the parametric curved surface in each of the divided regions are u and v, the value of the parameter u or v of the measurement point on the curve for parameter setting designated by the curve designating means C2911 is set. With respect to the points changed to the parameter setting means 2913 for assigning parameters to the respective measurement points so as to be aligned, and the direction in which the nodes and control points are added in each of the divided areas with respect to the automatic node determination means 2414 in FIG. Automatic node determination means 291 for automatically determining the position using the variance of the previous approximation error 24 and the parametric curved surface approximating means 2415 in FIG. 24, the parametric curved surface approximating means for sequentially approximating the three-dimensional coordinate value data of each area to the parametric curved surface using the nodes in each of the divided areas. 2915 and the data format conversion means 2417 in FIG. 24, each of the divided areas is a parametric curved surface approximation means 2915.
Data format conversion means 291 for converting the data compressed by the data format of the input destination system
This is a point changed to 7.

The operation of the three-dimensional shape input device configured as described above will be described. X-axis moving mechanism 290
An object to be measured 2904 is installed on the No. 5. Laser light source 290
The slit light 2903 from No. 1 has its optical path changed by the mirror 2902, and the object to be measured 2904 on the X-axis moving mechanism 2905.
Is irradiated. Hereinafter, the same operation as that of the first embodiment is performed until the entire three-dimensional shape data of the DUT 2904 is acquired. The curve designating means C2911 automatically extracts a curve that divides the shape of the measured object into areas using the three-dimensional coordinate value data representing the measured overall shape of the measured object, and further sets a threshold value for each area. By loosening and performing the same operation, a curve for setting parameters when approximating each area to a curved surface is automatically extracted. Thereafter, similarly to the fifth embodiment, the area dividing means 2912 divides the three-dimensional coordinate value data into a plurality of areas by the curve for area division extracted by the curve designating means C2911, and the parameter setting means 2913 and the automatic node conversion. The determining means 2914 and the parametric curved surface approximating means 2915 compress the shape information by approximating each area extracted by the area dividing means 2912 individually to a parametric curved surface in the same manner as in the second embodiment.
The data format converting means 2917 is the object to be measured 290 approximated by the parametric curved surface approximating means 2915.
The parameter values of a plurality of parametric curved surfaces representing the shape of the entire 4 are converted into a data format for the CAD system of the input destination, and the format-converted 3D shape data 2918 is input to the 3D CAD system 2919.

A more detailed operation of the curve designating means C2911 will be described. The curve designating means C2911 automatically extracts a curve that divides the shape of the measured object into areas using the three-dimensional coordinate value data representing the measured overall shape of the measured object, and further sets a threshold value for each area. By loosening and performing the same operation, a curve for setting parameters when approximating each area to a curved surface is automatically extracted. FIG. 30 shows a flow of a method for automatically dividing the object to be measured into areas by using the three-dimensional coordinate value data representing the entire shape of the object to be measured in the curve designating means C2911.

First, the three-dimensional coordinate values of the object measured in step 3001 are input. In step 3002, the measurement point sequence is divided into 3 × 3 small areas. In step 3003, a least squares approximation is performed to the quadratic polynomial (Equation 5) using a 5 × 5 point sequence in which each small area is further increased by one round.

[0088]

[Equation 5]

In step 3004, the average curvature H and the Gaussian curvature K in each small area are obtained by using (Equation 5) and (Equation 6).

[0090]

[Equation 6]

In step 3005, the average curvature H and the Gaussian curvature are
Label each subregion with a combination of K zero, positive, and negative.
Create a label map. Labels should be close to zero
Threshold value K to consider as zero₀, H₀(K₀> 0, H₀>
0) is set and K> K₀, H <-H₀Label 1, K
> K₀, H> H₀Label 2, K₀> K> -K₀, H <
ー H ₀Label 3, K₀> K> -K₀, H> H₀When
Label 4, K₀> K> -K₀, H₀> H> -H₀When
Le 5, K <-K₀, H₀> H> -H₀Label 6, K
<-K₀, H <-H₀When, label 7, K <-K₀, H>
H₀Then label 8 is given. Each label is shown in FIG. 31.
8 types of curved surfaces are shown. Label in step 3006
Pay attention to a certain small area on the map, and
Label of the small area that focuses on the highest number of label
Rewrite the bell number. If you have the same number of labels,
The same label in 12 neighborhoods, 15 neighborhoods, and a wide neighborhood
Count the number of numbers and rewrite the label of the small area of interest
It Next, in step 3007, the normal vector of each small area is calculated.
It is calculated using (Equation 5). 3 × 3 small each in step 3008
In the area, the small areas in the vicinity of 8 and the normal vector
Find the inner product of Le. In step 3009, the inner product value is binarized,
Get the roof edge. Step 3010: Align the roof edge
Consolidate small areas with the same label number to avoid cutting
At 3011, it is extracted as a closed region. 3 by the above operation
Automatically segment the area of a measured object from dimensional data
You can The curve to be obtained becomes the boundary line of the area.

Similarly, the three-dimensional coordinate value data input in the procedure 3001 is changed to the three-dimensional coordinate value data of each area in place of the three-dimensional coordinate value data of the entire object to be measured, and further, in Step 3009. A curve for parameter setting can be extracted by loosening the threshold value by digitization.

In the curve designating means C2911,
The method of designating a curve for parameter setting is such that the reflectance for slit light is different from the surface color of the DUT, the curve is specified on the DUT, and the brightness value is different in the image signal. A curve indicating means A2411 for extracting a portion as the curve, or a curve whose reflectance with respect to slit light is different from the surface color of the object to be measured is used to roughly specify the curve on the object to be measured, and the curve of the image signal Among them, a curve indicating means B2511 for extracting a curve based on the edge-likeness obtained by using the three-dimensional coordinate value data from a portion having a different brightness value and its peripheral portion, or a camera reads slit scattered light and projects slit light. An image of the entire measured object in a state in which the slit light is not captured, or the entire measured object that is not projecting the slit light is captured. The memory switching means for switching the output depending on whether the degree information is imaged, the first image memory for storing the image signal of the slit scattered light by switching the memory switching means, and the slit light is projected by switching the memory switching means. A second image memory for storing the whole luminance image of the measured object which is not present is provided separately, and the obtained curve is regarded as an edge portion of the measured object, and the luminance image is automatically obtained without human intervention. To determine the edge, or to specify the curve on the DUT using a color different from the surface color of the DUT and measure the condition when the camera reads the slit scattered light and does not project the slit light. A color camera that images the entire object, and the brightness information of the entire object to be measured in which the camera images the slit light or does not project the slit light. A memory switching unit that switches output depending on whether an image is captured, a first image memory that stores an image signal of slit scattered light by switching the memory switching unit, and the object that does not project slit light by switching the memory switching unit. A second image memory for storing the brightness image of the entire measured object is provided separately,
Extract the curve using the color difference from the luminance image, or
Alternatively, the measured three-dimensional coordinate value data may be displayed on a computer, and a human may use the mouse to indicate a curve for the displayed three-dimensional coordinate value data.

As described above, according to the present embodiment, the laser light source 2901 as the slit light source for generating the laser slit light for irradiating the DUT 2904 and the mirror 2902.
An X-axis moving mechanism 2905 that moves the device under test 2904 at a constant pitch in the X axis direction, a camera 2906 that captures the scattered light of the slit light by the device under test 2904, and the output signals from the camera 2906 are digitized. A / D converter 2907 for converting into an image signal, image memory 2908 for storing the image signal, and slit scattered light center position detecting means for detecting the center position of slit scattered light from the image signal stored in the image memory 2908. 2909, coordinate calculation means 2910 for calculating a three-dimensional coordinate value of the object to be measured 2904 from the slit scattered light center position, and a curve to be obtained is regarded as an edge portion of the object to be measured, and the three-dimensional coordinate value data is automatically acquired. The curve for segmentation is extracted by finding the edges using the Curve indicating means C2911 for extracting a curve for setting, area dividing means 2912 for dividing the three-dimensional coordinate value data into areas according to the curve for area division designated by the curve indicating means C2911, and each of the divided areas. In this case, if the parameters of the parametric curved surface are u and v, the parameters are set to the respective measurement points so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by the curve designating means C2911 are aligned. A parameter setting means 2913 to be given, a nodal point automatic determining means 2914 for automatically determining the direction and position of the nodal point and control point to be added in each of the divided areas using the previous approximation error variance, The three-dimensional coordinate value data of each area is successively parametrically curved by using the nodes in each of the created areas. A parametric curved surface approximating means 2915 for approximating, a data format converting means 2917 for converting the data compressed by the parametric curved surface approximating means 2915 for each of the divided areas into a data format of the input destination system, and a scanner control for controlling the entire system. By providing the means 2916, the amount of data can be efficiently reduced while the measured shape is saved, and the intended shape can be easily changed or modified in the input destination system in a three-dimensional CAD or computer graphic format. System 2919.

(Embodiment 8) Hereinafter, an eighth embodiment of the present invention will be described with reference to the drawings.

FIG. 32 shows 3 in the eighth embodiment of the present invention.
It is a block diagram of a three-dimensional shape input device. In FIG. 32, 3
Reference numeral 201 denotes a laser light source for generating slit light for irradiating the DUT 3204, 3202 a mirror for changing the optical path of the slit light, 3203 for slit light irradiating the DUT 3204, 3205 for the X-axis direction of the DUT 3204. An X-axis moving mechanism for moving the slit light 3203 at a constant pitch, a camera for reading the scattered light of the slit light 3203, and an image of the entire measured object 3204 in a state where the slit light is not projected, 3207 are digital output signals of the camera 3206. The A / D converter 3208 for converting into a converted image signal switches the input memory depending on whether the camera 3206 has imaged the slit light or the entire brightness information of the DUT 3204 not projecting the slit light. Memory switching means 3209 is an image memory for storing the image signal of the slit scattered light. Li, 3210 an image memory for storing the luminance image of the entire object to be measured not by projecting slit light, 3211 memory switching means 3208, an image memory 3
209 and 3210 control CPU, 3212 slit slit light center position detection means for calculating the center position of slit scattered light from the image signal from the image memory 3209, 3
213 is a coordinate calculation means for calculating a three-dimensional coordinate value from the center position of the slit scattered light, and 3214 is an image memory 3210.
The curve indicating means D and 3215 for extracting the curve for automatically dividing the area and the curve for the parameter setting means from the brightness image of the measured object stored in the area dividing area extracted by the curve indicating means D3214. Area dividing means for dividing the entire shape of the object to be measured into a plurality of areas using a curve for, and 3216 on the continuous curves extracted by the curve indicating means D3214, where u and v are parameters of the parametric curved surface. Parameter setting means for assigning a parameter to each measurement point so that the u or v values of the measurement points are made uniform, and 3217 is the direction and position at which the node and the control point are added, using the variance of the previous approximation error. A node automatic determination means for automatically determining, 3218 is three-dimensional using the nodes for each area divided by the area dividing means 3215 Parametric curved surface approximating means for approximating the standard value data to the parametric curved surface, 3220 is data format converting means for converting the data compressed by the parametric curved surface approximating means 3218 into the data format of the input destination system, and 3219 is the scanner control for controlling the entire system. Means, 3221 is three-dimensional shape data obtained by this apparatus, and 3222 is a three-dimensional CAD system.

The operation of the three-dimensional shape input device configured as described above will be described. First, the laser light source 3
The slit light 3203 from 201 has its optical path changed by the mirror 3202 and is irradiated on the DUT 3204. The DUT 3204 moves in the X-axis direction at a constant pitch by the X-axis moving mechanism 3205 while receiving the slit light 3203.
The camera 3206 captures the slit scattered light from the moving object 3204 in synchronization with the movement. In this case, the captured image of the slit scattered light on the camera 3206 shows the unevenness of the slit scattered light according to the shape of the surface of the DUT 3204. The A / D converter 3207 is a camera 3206.
Signal is converted into a digital signal, and this signal is temporarily stored in the image memory 3209 as an image signal by the memory switching means 3208. The slit scattered light center position detecting means 3212 finds the center position of the luminance of the slit scattered light for each horizontal scanning line of the camera from the image signal and regards it as the image of the measurement point. In the coordinate calculation means 3213, the angle formed by the straight line connecting the image of the measuring point and the lens center of the camera and the optical axis of the camera,
Then, the three-dimensional coordinate value of the measurement point is calculated by the principle of the triangulation method using the baseline length (distance from the slit light source to the camera lens center) and the projection angle of the laser slit light. In this way, the three-dimensional shape data of the measured object is acquired for one slit light. Then, by acquiring the three-dimensional shape data for each moving pitch, the three-dimensional shape data of the entire measured object 3204 is finally acquired.

Next, the projection of the slit light on the object to be measured 3204 is stopped, and the camera 3206 again sets the object to be measured 320.
4 is imaged. In this case, the luminance information of the entire measured object 3204 is imaged on the camera 3206. The A / D converter 3207 converts the signal from the camera 3206 into a digital signal, and this signal is stored in the image memory 3210 as an image signal by the memory switching means 3208. The curve designating means D3213 automatically extracts from the image signal a curve for dividing the shape of the DUT 3204 and a curve used for parameter setting when the respective surfaces are approximated to a curved surface. Thereafter, similarly to the fifth embodiment, the area dividing means 3215 divides the three-dimensional coordinate value data into a plurality of areas by the curve for area division extracted by the curve designating means D3214, and the parameter setting means 3216 and node automatic. Deciding means 3217,
The parametric curved surface approximating means 3218 is the area dividing means 3
The shape information is compressed by individually approximating each region extracted by 215 to a parametric curved surface in the same manner as in the second embodiment, and the data format conversion means 3220 is the whole measured object 3204 approximated by the parametric curved surface approximating means 3218. The parameter values of a plurality of parametric curved surfaces representing the shape of are converted into the data format for the CAD system of the input destination, and the format is converted.
The three-dimensional shape data 3221 is the three-dimensional CAD system 322.
Entered in 2.

The detailed operation of the curve designating means D3214 will be described below. FIG. 33 shows a curve designating means 3214.
2 shows a flow of a method for automatically segmenting an object to be measured using luminance information.

First, in step 3301, the brightness information of the object to be measured is input. Kirsch is performed on the luminance image in step 3302.
Apply the type operator. FIG. 27 shows a Kirsch type operator. In step 3303, the first-order difference value by the Kirsch type operator is binarized by the preset threshold value, and the edge is extracted. In Step 3304, Step 3303
4 edge expansion is applied to the edge image obtained in (4). Step 330
At 5, the edge is thinned. The closed region divided by the edge is extracted by tracing the edge in step 3306. In step 3307, the closed areas are integrated. In the integration, R is calculated by (Equation 7) from the pixel number I of the common boundary between two adjacent closed areas and the pixel number W of which the density difference is smaller than the threshold value at the common boundary, and is set in advance. If R is larger than the existing threshold R ₀ , the two areas are integrated.

[0101]

[Equation 7]

With the above operation, the area of the object to be measured can be automatically divided from the luminance data. The curve that divides the obtained region indicates the boundary of the divided region. In the above procedure, the brightness image input in step 3301 is used as the brightness image in each divided area, and the threshold for binarization in step 3303 is relaxed to automatically extract a curve for parameter setting. be able to.

In the curve designating means D3214,
The method of designating a curve for parameter setting is such that the reflectance for slit light is different from the surface color of the DUT, the curve is specified on the DUT, and the brightness value is different in the image signal. A curve indicating means A2411 for extracting a portion as the curve or a color having a reflectance for slit light different from the surface color of the object to be measured is used to roughly and thickly specify the curve on the object to be measured. In the curve indicating means B2511 for extracting the curve based on the edge-likeness obtained by using the three-dimensional coordinate value data from the portion having different luminance values and the peripheral portion thereof, or the obtained curve is the edge portion of the object to be measured. Therefore, the curve indicating means C2911 for automatically obtaining an edge by using the three-dimensional coordinate value data without any human intervention, or a color different from the surface color of the object to be measured is used. A color camera that specifies the curved line on the DUT and images the entire DUT in a state in which the camera reads the slit scattered light and does not project the slit light. A memory switching unit for switching the output depending on whether or not the brightness information of the entire object to be measured, which is not projecting light, is switched; a first image memory for storing the image signal of the slit scattered light by switching the memory switching unit; A second image memory for storing a brightness image of the entire object to be measured which is not projecting slit light by switching the memory switching means is separately provided, and the curve is extracted from the brightness image by using a color difference. Alternatively, the measured three-dimensional coordinate value data is displayed on a computer, and a human is used for the displayed three-dimensional coordinate value data by using a mouse. It may instruct the curve.

As described above, according to this embodiment, the laser light source 3201 and the mirror 3202 as a laser light source for generating the slit light for irradiating the DUT 3204 and the DUT 3204 at a constant pitch in the X-axis direction. X to move
The axis moving mechanism 3205 and the DUT 320 in a state where the scattered light of the slit light is read and the slit light is not projected.
4. A camera 3206 that captures the entire image, an A / D converter 3207 that converts the output signal of the camera 3206 into a digitized image signal, and a measured object that does not project slit light or projects slit light with the camera 3206. A memory switching unit 3208 that switches the output depending on whether the brightness information of the entire object 3204 is imaged, a first image memory 3209 that stores the image signal of the slit scattered light, and the brightness of the entire measured object that does not project the slit light. A second image memory 3210 for storing images and a memory switching means 32
08 and CP for controlling the image memories 3209 and 3210
U3211, slit scattered light center position detecting means 3212 for calculating the central position of the slit scattered light from the image signal from the first image memory 3209, and coordinate calculating means for calculating a three-dimensional coordinate value from the central position of the slit scattered light. 32
13, a curve designating means D3214 for extracting a curve for performing area division and a curve for parameter setting from the luminance image of the DUT 3204 stored in the second image memory 3210, and a curve designating means D3214. Area dividing means 3215 for dividing the three-dimensional coordinate value data according to the indicated curve for area division, and curve indicating means D3214 when the parameters of the parametric curved surface in each of the divided areas are u and v The parameter setting means 3216 for assigning parameters to the respective measurement points so that the values of the parameters u or v of the measurement points on the curved line for the parameter setting are aligned, and the nodes in each of the divided areas. Automatic determination of the direction and position of the control point to be added automatically using the variance of the previous approximation error Means 3217, a parametric surface approximation means 3218 for sequentially approximating the three-dimensional coordinate value data of each area into a parametric surface using the nodes in each of the divided areas, and a parametric surface approximation for each of the divided areas. By providing the data format conversion means 3220 for converting the data compressed by the means 3218 into the data format of the input destination system and the scanner control means 3219 for controlling the entire system, the amount of data can be efficiently stored while keeping the measured shape. Can be input to the three-dimensional CAD or computer graphics system 3222 in a data format in which the number of characters is reduced and the intended shape can be easily changed or modified in the input destination system.

(Ninth Embodiment) A ninth embodiment of the present invention will be described below with reference to the drawings.

FIG. 34 shows a third embodiment of the present invention.
It is a block diagram of a three-dimensional shape input device. In FIG. 34, 3
Reference numeral 401 denotes a laser light source for generating slit light for irradiating the DUT 3404, 3402 for a mirror for changing the optical path of the slit light, 3403 for slit light for irradiating the DUT 3404, and 3405 for the X-axis direction of the DUT 3404. X-axis moving mechanism for moving the color camera 3406 at a constant pitch, 3407 is an A / D converter for converting the output signal of the color camera 3406 into a digitized image signal, and 3408 is the color camera 34.
A memory switching unit 3409 for switching the input memory depending on whether the slit light is imaged by 06 or the luminance information of the entire DUT 3404 not projecting the slit light is imaged for storing the image signal of the slit scattered light. A memory, 3410, is an image memory for storing a luminance image of the entire DUT that does not project slit light, 341.
1 is a memory switching unit 3408, image memories 3409, 3
CPU for controlling 410, 3412 for image memory 340
The slit scattered light center position detecting means for calculating the central position of the slit scattered light from the image signal from the reference numeral 9, reference numeral 3413 for coordinate calculating means for calculating a three-dimensional coordinate value from the central position of the slit scattered light, reference numeral 3415 for the curve indicating means E3414. A region dividing unit 341 that divides the entire shape of the measured object into a plurality of regions using the extracted curve for dividing the region.
A parameter 6 is given to each measurement point so that the parameter values u or v of the measurement points on the continuous curve extracted by the curve designating means E3414 are made uniform if the parameters of the parametric curved surface are u and v. Parameter setting means, 3417 is automatic node determining means for automatically determining the direction and position of addition of nodes and control points using the variance of the previous approximation error, and 3418 is each area divided by the area dividing means 3415. , Parametric curved surface approximating means for approximating the three-dimensional coordinate value data to a parametric curved surface using the nodes, 3420 is data format converting means for converting the data compressed by the parametric curved surface approximating means 3418 into the data format of the input destination system, 3419 Is a scanner control means for controlling the entire system, and 3421 is a main unit. 3-dimensional shape data obtained by, 3
Reference numeral 422 is a three-dimensional CAD system, and the above is the same as the configuration of FIG.

32 is different from the configuration of FIG. 32 in that the curve and each region for dividing the entire shape of the DUT 3204 of FIG. 32 by using a color different from the surface color of the DUT 3404 are used. Of the slit light 34 for the camera 3206 in FIG. 32 and the point where the curve used when setting parameters when approximating the
32 is changed to a color camera 3406 that captures the entire measured object 3404 in a state in which scattered light of 03 is not projected and slit light is not projected, and the curve indicating means D32 in FIG.
14, a curve designating means E3414 for extracting a curve for automatically dividing an area using a color difference from the color image of the DUT stored in the image memory 3410 and a curve for the parameter setting means. It is a point changed to.

The operation of the three-dimensional shape input device configured as described above will be described. First, the laser light source 3
The slit light 3403 from 401 changes the optical path by the mirror 3402 and irradiates the DUT 3404. The device under test 3404 moves in the X-axis direction at a constant pitch by the X-axis moving mechanism 3405 while receiving the slit light 3403. The color camera 3406 images the slit scattered light from the moving object under measurement 3404 in synchronization with the movement. in this case,
The image of the slit scattered light on the color camera 3406 that has been captured shows the unevenness of the slit scattered light according to the shape of the surface of the DUT 3404. The A / D converter 3407 converts the signal from the color camera 3406 into a digital signal, and the signal is converted by the memory switching means 3408 into the image memory 340.
9 is temporarily stored as an image signal. The slit scattered light center position detecting means 3412 finds the center position of the brightness of the slit scattered light for each horizontal scanning line of the color camera from the image signal and regards it as an image of the measurement point. Coordinate calculation means 3413
At the angle between the straight line connecting the image of the measurement point and the lens center of the color camera and the optical axis of the color camera, and the baseline length (distance from the slit light source to the camera lens center) and the projection angle of the laser slit light. The three-dimensional coordinate value of the measurement point is calculated by the principle of triangulation method. In this way, the three-dimensional shape data of the measured object is acquired for one slit light. Then, by acquiring the three-dimensional shape data for each movement pitch, finally, the total 3
Obtains dimensional shape data.

Next, the projection of the slit light on the object to be measured 3404 is stopped, and the color camera 3406 takes an image of the object to be measured 3404 again. In this case, the color camera 3406
Luminance information of the entire measured object 3404 is imaged on the upper side.
The A / D converter 3407 converts the signal from the color camera 3406 into a digital signal, and this signal is stored in the memory switching means 34.
08, the image signal is stored in the image memory 3410 as an image signal. The curve designating means E3414 automatically extracts a curve for dividing the shape of the DUT 3404 from the image signal and a curve used for parameter setting when the respective areas are approximated to a curved surface by using the color difference. Thereafter, similarly to the fifth embodiment, the area dividing means 3415 divides the three-dimensional coordinate value data into a plurality of areas by the curve for area division extracted by the curve designating means E3414, and the parameter setting means 3416, the automatic node conversion. The determining means 3417 and the parametric curved surface approximating means 3418 compress the shape information by approximating each area extracted by the area dividing means 3415 individually to a parametric curved surface in the same manner as in the second embodiment.
The data format converting means 3420 is the object to be measured 340 approximated by the parametric curved surface approximating means 3418.
The parameter values of a plurality of parametric curved surfaces representing the shape of the entire 4 are converted into the data format for the CAD system of the input destination, and the format-converted 3D shape data 3421 is input to the 3D CAD system 3422.

The more detailed operation of the curve designating means E3414 will be described below. FIG. 35 shows a curve designating means E341.
4 shows a flow of a method of automatically dividing the object to be measured into areas using the luminance information.

First, in step 3501, the color information of the object to be measured is input. In step 3502, the (R, G, B) values of each pixel of the color image are in the uniform color space (H, V, C).
Convert to. In step 3503, (H, V, C) of each pixel
The value of is voted in the HVC Munsell color space, which is the feature space. In step 3504, color space clustering is performed. Since the surface color of the DUT 3404, the color of the curve for dividing the region, and the color of the curve for setting the parameters are different from each other, three clusters are considered. First, three arbitrary representative vectors are taken, and they are regarded as the standard pattern of each cluster C1, C2, C3. Next, the distance between a certain vector p _i in the space and the standard patterns C1, C2, C3 of the clusters is obtained, and the vector p _i belongs to the cluster having the smallest distance value. The above operation is performed for all vectors p _i (i = 1,
..., n). Next, in each cluster, the average vector is set as a new standard pattern, and a certain vector p _i in the space and standard patterns C1 and C
2, the distance from C3 is calculated, and the operation of making the vector p _i belong to the cluster having the smallest distance value is performed on all the vectors p _i.
Repeat for (i = 1, ..., N). C1, C2, C3 when the standard pattern of each cluster becomes equal to the previous standard pattern are obtained as the result of clustering. In step 3505, the three clusters are labeled c
Given ₁ , c ₂ and c ₃ , create a label map. Step 3
If the cluster having the largest number of vectors belonging to the cluster out of the three clusters is c _{1 in} 506, it is determined that c ₁ is the surface color of the DUT 3404. In step 3507, pixels having labels c ₂ and c ₃ in the label map are tracked, and the closed one of c ₂ and c ₃ is determined to be a curve for area division. For example, each curve can be extracted as a set of pixels having a label c ₂ for a curve for area division and as a set of pixels having a label c ₃ for a parameter setting. Note that which cluster represents which curve can be easily determined by the position of the cluster in the Munsell color space by knowing the color of the curve in advance.
Although the Munsell color space is used as the feature space for color-based region division, other color spaces may be used and the present invention is not limited thereto.

In the curve designating means E3414,
The method of designating a curve for parameter setting is such that the reflectance for slit light is different from the surface color of the DUT, the curve is specified on the DUT, and the brightness value is different in the image signal. A curve indicating means A2411 for extracting a portion as the curve or a color having a reflectance for slit light different from the surface color of the object to be measured is used to roughly and thickly specify the curve on the object to be measured. The curve indicating means B2511 for extracting the curve based on the edge-likeness obtained by using the three-dimensional coordinate value data from the portion having different luminance values and the peripheral portion thereof, or the obtained curve is the edge portion of the object to be measured. It is assumed that the curved line indicating means C2911 for automatically obtaining the edge using the three-dimensional coordinate value data without any human intervention or the camera reads the slit scattered light. A camera that images the entire object to be measured in the state where the slit light is not projected, the camera images the slit light, or images the brightness information of the entire object to be measured that does not project the slit light. The memory switching means for switching the output depending on whether or not, the first image memory for storing the image signal of the slit scattered light by switching the memory switching means, and the DUT that does not project the slit light by switching the memory switching means. A second image memory for storing the entire brightness image of
A curve designating means D3214 for automatically extracting the curve from the brightness image, or the measured three-dimensional coordinate value data is displayed on a computer, and a human is used to curve the displayed three-dimensional coordinate value data using a mouse. May be instructed.

As described above, according to this embodiment, the laser light source 3401 and the mirror 3402 as the laser light source for generating the slit light for irradiating the DUT 3404 and the DUT 3404 at a constant pitch in the X-axis direction. X to move
The axis moving mechanism 3405 and the DUT 340 in a state where the scattered light of the slit light is read and the slit light is not projected.
4. A color camera 3406 that captures the entire image, an A / D converter 3407 that converts the output signal of the color camera 3406 into a digitized image signal, and the slit light is captured by the color camera 3406 or the slit light is projected. Memory switching means 3408 for switching the output depending on whether or not the brightness information of the entire non-measured object 3404 is captured.
A first image memory 3409 for storing the image signal of the slit scattered light, and a second image memory 3410 for storing the brightness image of the entire DUT which does not project the slit light.
A CPU 3411 for controlling the memory switching means 3408 and the image memories 3409 and 3410; and a slit scattered light center position detecting means 341 for calculating the central position of the slit scattered light based on the image signal from the first image memory 3409.
2, coordinate calculation means 3413 for calculating a three-dimensional coordinate value from the center position of the slit scattered light, and the second image memory 34.
A curve indicating means E3414 for extracting a curve for performing area division and a curve for parameter setting from the color image of the DUT 3404 stored in 10, and for area division indicated by the curve indicating means E3414. Area dividing means 3415 for dividing the three-dimensional coordinate value data into areas according to a curve, and a curve for parameter setting designated by the curve designating means E3414, where u and v are parameters of the parametric curved surface in each of the divided areas. Parameter setting means 3416 for giving a parameter to each measurement point so that the value of the parameter u or v of the measurement point on the top is made uniform, and the direction and position of adding the node and the control point in each of the divided areas. Is automatically determined using the variance of the previous approximation error.
17, and parametric curved surface approximating means 3418 for sequentially approximating the three-dimensional coordinate value data of each area to the parametric curved surface using the nodes in each of the divided areas.
By providing data format conversion means 3420 for converting the data compressed by the parametric curved surface approximation means 3418 into the data format of the input destination system, and scanner control means 3419 for controlling the entire system, Input data to the 3D CAD or computer graphics system 3422 in a data format that can efficiently reduce the amount of data while saving the measured shape, and can easily change or correct the intended shape in the input destination system. You can

(Embodiment 10) Hereinafter, a tenth embodiment of the present invention will be described with reference to the drawings. FIG. 36 is a block diagram of a three-dimensional shape input device according to the tenth embodiment of the present invention.

In FIG. 36, reference numeral 3601 denotes the object to be measured 36.
Laser light source for generating slit light for irradiation of 04, 3
Reference numeral 602 is a mirror for changing the optical path of the slit light, 360
3 is a slit light irradiating the DUT 3604, 3605
Is an X-axis moving mechanism that moves the DUT 3604 at a constant pitch in the X-axis direction, 3606 is a camera that reads the scattered light of the slit light 3603, and 3607 is A / that converts the output signal of the camera 3606 into a digitized image signal. D converter, 3608 is an image memory for storing image signals, 360
Reference numeral 9 is a slit scattered light center position detecting means for calculating the central position of the slit scattered light from the image signal, 3610 is a coordinate calculating means for calculating a three-dimensional coordinate value from the central position of the slit scattered light, and 3616 is a scanner for controlling the entire system. Control means, 3618 is three-dimensional shape data obtained by this device,
619 is a three-dimensional CAD system, and the above is the same as the configuration of FIG.

The difference from the configuration of FIG. 24 is that the measured three-dimensional coordinate value data for the curve designating means A 2411 in FIG. 24 is measured with respect to the three-dimensional coordinate value data displayed on the three-dimensional CAD system 3619. A point for changing a curve for dividing the shape of the object 3604 and a curve used for parameter setting when approximating each area into a curved surface to a curve indicating means F3611 specified by a human using the mouse 3620, The point at which the mouse 3620 to be used is added, at the time when the three-dimensional coordinate value data is obtained in order to indicate a curve instead of simply using the three-dimensional CAD system as an output destination for the three-dimensional CAD system 2419 in FIG. Data with 3D CAD system 3619
24. The area changed to the three-dimensional CAD system 3619 displayed in FIG. 24 and the area dividing means 2412 in FIG. 24 are divided into three areas according to the curve for area division designated by the curve designating means F. With respect to the point changed to the means 3612 and the parameter setting means 2413 in FIG. 24, if the parameters of the parametric curved surface are u and v in each of the divided areas, the parameter setting means F3611 for setting the parameters is set. The point is changed to the parameter setting means 3613 for assigning a parameter to each measurement point so that the values of the parameter u or v of the measurement point on the curve are made uniform, and the node automatic determination means 2414 in FIG. 24 is divided. In each of the areas, the direction and position of adding the node and control point are set to the previous approximation error. For the points changed to the automatic node determination means 3614 which is automatically determined by using the dispersion and the parametric curved surface approximating means 2415 in FIG. With respect to the parametric curved surface approximating means 3615 for sequentially approximating the coordinate value data to the parametric curved surface, and the data format converting means 2417 in FIG.
Data format conversion means 361 for converting the data compressed by the data into the data format of the input destination system
This is a point changed to 7.

The operation of the three-dimensional shape input device configured as described above will be described. First, the DUT 3604 is installed on the X-axis moving mechanism 3605. The slit light 3603 from the laser light source 3601 has its optical path changed by the mirror 3602, and the object to be measured 3 on the X-axis moving mechanism 3605.
Irradiation 604. Hereinafter, the same operation as that of the first embodiment is performed until the entire three-dimensional shape data of the DUT 3604 is acquired. The curve designating means F3611 forms a triangular patch by using the three-dimensional coordinate value data representing the measured overall shape of the measured object to form a surface on the point sequence and the three-dimensional CA.
Display on D system 3619. And three-dimensional CAD
With respect to the measurement shape of the DUT 3604 displayed on the system 3619, a human is instructed by clicking the position of the curve for dividing the shape of the DUT 3604 with the mouse 3620. In the same manner, next, by clicking another button of the mouse 3620, a person specifies a curve to be used for parameter setting when each area is approximated to a curved surface.
Thereafter, similarly to the fifth embodiment, the area dividing means 3612 divides the three-dimensional coordinate value data into a plurality of areas by the curve for area division extracted by the curve designating means F3611, and the parameter setting means 3613 and the automatic node conversion. The determining means 3614 and the parametric curved surface approximating means 3615 compress the shape information by approximating each area extracted by the area dividing means 3612 to a parametric curved surface individually in the same manner as in the second embodiment, and the data format converting means 3617 is parametric. The parameter values of a plurality of parametric curved surfaces representing the overall shape of the DUT 3604 approximated by the curved surface approximating means 3615 are converted into a data format for the CAD system of the input destination, and the format-converted three-dimensional shape data 3618 is three-dimensional. Entered CAD system 3619 It is. In this embodiment, three-dimensional CAD
Mouse 3620 when designating a curve with system 3619
However, different indicating devices may be used. Also,
The curve designating means F3611 uses the three-dimensional coordinate value data representing the overall shape of the measured object to be displayed on the three-dimensional CAD system 3619 as a gradation corresponding to the z coordinate value, and divides the shape of the object to be measured 3604. By clicking with the mouse 3620 the position of the curve for
Similarly, a human may indicate a curve to be used for parameter setting when each area is approximated to a curved surface by clicking another button of the mouse 3620.

In the curve designating means F3611,
The method of designating a curve for parameter setting is such that the reflectance for slit light is different from the surface color of the DUT, the curve is specified on the DUT, and the brightness value is different in the image signal. A curve indicating means A2411 for extracting a portion as the curve or a color having a reflectance for slit light different from the surface color of the object to be measured is used to roughly and thickly specify the curve on the object to be measured. The curve indicating means B2511 for extracting the curve based on the edge-likeness obtained by using the three-dimensional coordinate value data from the portion having different luminance values and the peripheral portion thereof, or the obtained curve is the edge portion of the object to be measured. It is assumed that the curved line indicating means C2911 for automatically obtaining the edge using the three-dimensional coordinate value data without any human intervention or the camera reads the slit scattered light. The entire object to be measured in the state where the slit light is not projected is imaged, and the camera images the slit light or the brightness information of the entire object to be measured in which the slit light is not projected. The memory switching means for switching the output depending on whether or not, the first image memory for storing the image signal of the slit scattered light by switching the memory switching means, and the DUT that does not project the slit light by switching the memory switching means. A second image memory for storing the entire luminance image of the above is provided separately, and the obtained curve is regarded as the edge portion of the object to be measured, and the edge is automatically detected using the luminance image without human intervention. The curve is specified on the object to be measured using the curve indicating means D3214 to be obtained or a color different from the surface color of the object to be measured, and the camera reads the slit scattered light. In addition, with a color camera that images the entire DUT without projecting the slit light, the camera captures the slit light or displays the brightness information of the entire DUT that does not project the slit light. A memory switching means for switching the output depending on whether an image is taken, and a first image memory for storing the image signal of the slit scattered light by switching the memory switching means,
A second image memory for storing a brightness image of the entire object to be measured on which slit light is not projected by switching the memory switching means is separately provided, and the curve is extracted from a color image using a color difference. The curve designating means E3414 may be used.

As described above, according to this embodiment, two colors having different reflectances for slit light from the surface color of the object to be measured and having different reflectances to each other are used, and the object to be measured is respectively colored. An object to be measured 3604 in which a person roughly and roughly entered a curve for dividing an area for performing approximation by dividing into a plurality of areas and a curve for setting parameters when approximating each area to a curved surface, A laser light source 3601 and a mirror 3602 as a slit light source for generating a laser slit light for irradiating a measured object 3604, and an object to be measured 3604.
X-axis movement mechanism 36 for moving the X-axis direction at a constant pitch
05, a camera 3606 that captures the scattered light of the slit light by the DUT 3604, an A / D converter 3607 that converts the output signal from the camera 3606 into a digitized image signal, and an image that stores the image signal. Memory 360
8 and the slit scattered light center position detecting means 3609 for detecting the center position of the slit scattered light from the image signal stored in the image memory 3608, and the three-dimensional coordinate value of the DUT 3604 is calculated from the slit scattered light center position. Coordinate calculation means 3610, the measured three-dimensional coordinate value data are displayed on the three-dimensional CAD system 3619, and a curve and each curve for dividing the shape of the DUT 3604 with respect to the displayed three-dimensional coordinate value data. A curve designating means F3611 that a human uses a mouse 3620 to designate a curve used for parameter setting when a region is approximated to a curved surface, a mouse 3620 used to designate a curve, and a three-dimensional CAD system are simply used as output destinations. 3D CAD system 3 when the 3D coordinate value data is obtained to indicate a curve instead of 19, a three-dimensional CAD system 3619 displayed on the screen 19, area dividing means 3612 which divides the three-dimensional coordinate value data into areas according to a curve for area division designated by the curve designating means F3611, and divided areas. If the parameters of the parametric curved surface are u and v, the parameters are given to the respective measurement points so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by the curve designating means 3611 are aligned. Parameter setting means 3613, automatic nodal point determining means 3614 for automatically determining the direction and position of addition of nodal points and control points in each of the divided areas by using the variance of the previous approximation error. The three-dimensional coordinate value data of each area is successively approximated to a parametric curved surface by using the nodes in each area. Parametric curved surface approximating means 3615, data format converting means 3617 for converting the data compressed by the parametric curved surface approximating means 3615 into the data format of the input destination system, and scanner controlling means for controlling the entire system. By providing 3616,
The amount of data is efficiently reduced while saving the measured shape, and the intended shape change in the input destination system,
Data can be input to the three-dimensional CAD or computer graphics system 3619 in a data format that can be easily corrected.

(Eleventh Embodiment) The eleventh embodiment of the present invention will be described below with reference to the drawings. FIG. 37 is a block diagram of a three-dimensional shape input device according to the eleventh embodiment of the present invention.

In FIG. 37, reference numeral 3701 denotes the object to be measured 37.
A laser light source that generates a slit light to irradiate 04,
The object to be measured 3704 is a curve that divides the shape of the object to be measured 3704 into regions and each region by using two colors having different surface colors of the object to be measured and reflectances for slit light and different reflectances from each other. The curve used for parameter setting when approximating the curved surface is roughly and thickly written by a person. 3702 is a mirror for changing the optical path of slit light,
Reference numeral 3703 denotes slit light for irradiating the DUT 3704, 3
705 is an X-axis moving mechanism that moves the DUT 3704 in the X-axis direction at a constant pitch, and 3706 is slit light 3703.
A camera for reading scattered light from the camera 3707 is a camera 3706.
A / D that converts the output signal of the device into a digital image signal
A converter 3708 is an image memory for storing image signals,
Reference numeral 709 is a slit scattered light center position detecting means for calculating the central position of the slit scattered light from the image signal, 3710 is a coordinate calculating means for calculating a three-dimensional coordinate value from the central position of the slit scattered light, and 3711 is written in the measured object 3704. Curve extracting means B, 371 for extracting a curve for dividing the region based on the edge-likeness obtained from the three-dimensional coordinate value data from the two types of curved lines thus formed and the curve used for parameter setting.
2 is an area dividing means for dividing the three-dimensional coordinate value data into areas according to the curve for area division designated by the curve designating means 3711B, and 3713 is the parametric curved surface parameters u and v in each of the divided areas. Parameter setting means for assigning a parameter to each measurement point so that the values of the parameter u or v of the measurement points on the curve for parameter setting designated by the curve designating means B3711 are aligned, and 3714 for the divided areas. A node automatic determining means for automatically determining the direction and position of the addition of the node and the control point in each of them using the variance of the previous approximation error, 3715 uses the node in each of the divided areas. Parametric curved surface approximating means for sequentially approximating the three-dimensional coordinate value data of each region to a parametric curved surface, 3717 is division Each of the divided areas converts the data compressed by the parametric curved surface approximating means 3715 into the data format of the input destination system, 3716 is the scanner controlling means for controlling the entire system, and 3718 is the device obtained by this apparatus. Dimensional shape data, 3719, is a three-dimensional CAD system, and the above is the same as the configuration of FIG.

The difference from the configuration of FIG. 25 is that there is a boundary discriminating means 3720 for discriminating whether the boundary of the area is a roof edge, a smooth edge having a positive radius (convex curved surface) or a smooth edge having an inverse radius (concave curved surface), and A rounding radius estimating means 3721 for estimating rounding radii of rounded and inverse rounded lines, and data of the rounded part is deleted from the measured data region in the regular rounded part to open an interval between the regions, and a differential value is continuous in the reverse rounded part. Area type generation unit 37 for each type that expands the measurement data of the internal area to generate an area
22 is newly provided.

The operation of the three-dimensional shape input device configured as described above will be described. First, the DUT 37
The whole three-dimensional shape data of No. 04 is acquired, and it is used for setting the curve and the parameter for dividing the area from the curve written on the DUT 3704 based on the edge-likeness obtained from the measurement data by the curve extracting unit B3711. A curve is extracted, and the area dividing means 3712 causes the curve designating means B371.
The operation is the same as that of the sixth embodiment until each region is divided based on the curve extracted in 1.

The boundary discriminating means 3720 discriminates whether the boundary of the region is the roof edge, the smooth edge having the normal radius or the smooth edge having the inverse radius. Rounding radius estimation means 3
When it is determined that the boundary is a smooth edge, the reference numeral 721 estimates the rounding radius of the normal radius or the inverse radius. The radius-type-specific region boundary generation means 3722 expands the measurement data of the internal region so that the rounded part is deleted from the measurement data region in the regular radius part to leave a space between the regions, and the differential value is continuous in the inverse radius part. To regenerate the area.

FIG. 38 (a) shows a cross-sectional view near the boundary of the area in the case of the positive radius, and FIG. 38 (b) shows a cross-sectional view near the boundary of the area in the case of the reverse radius. By inputting to CAD with a surface having such a boundary, CAD
After inputting, it becomes easier to fillet in CAD. In particular, in the joint portion of the surfaces, corrections and the like are likely to occur after inputting into CAD, so it is necessary to create a boundary that can cope with such an arbitrary rounding radius.

In the same way as in the sixth embodiment, the parameter setting means 3713, the automatic node determination means 3714, and the parametric curved surface approximation means 3715 are applied to the respective areas recreated by the area type-specific area boundary generation means 3722. Similar to 2, the shape information is compressed by approximating the parametric curved surface, and the data format converting means 3717 inputs the parameter values of a plurality of parametric curved surfaces representing the shape of the whole measured object approximated by the parametric curved surface approximating means 3715. The three-dimensional shape data 3718 converted into the data format for the previous CAD system and the format converted is input to the three-dimensional CAD system 3719.

The detailed operation of the boundary discriminating means 3720 will be described below. FIG. 39 shows the boundary discriminating means 372.
0 shows a flow of a method for discriminating whether the boundary of one region is a roof edge, a smooth edge having a positive radius or a smooth edge having an inverse radius.

First, 3 of the point sequence measured in step 3901
Enter the dimensional coordinate value data. In step 3902, the boundary information of one area is input. In step 3903, the first-order differential operator is applied to the z coordinate value of one measurement point on the boundary. FIG. 27 shows a Kirsch operator as a first-order differential operator. Step 3
At 904, the operator selected at step 3903 is applied to the z coordinate values of all the measurement point sequences inside the boundary. In step 3905, the first-order difference value obtained in step 3904 is again set to Ki.
By applying the operator of the rsch operator, it is possible to obtain the secondary difference value in the direction perpendicular to the region boundary at each measurement point. In step 3906, the first-order difference value obtained in step 3904 in the measurement point sequence aligned in the vertical direction on the area boundary is approximated to (Equation 8) by least squares.

[0129]

[Equation 8]

In step 3907, the second-order difference value obtained in step 3905 in the sequence of measurement points arranged in the vertical direction on the area boundary is approximated to (Equation 9) by least squares.

[0131]

[Equation 9]

Procedure 3908 is a₁And a₂Bounded by the sign of
Label the top measurement points by shape.
A for labeling₁And a₂For a threshold value a close to zero
_TenAnd a ₂₀Is set in advance. However, a_Ten, A₂₀With
Is a positive value. And a₁> A_TenAnd a₂<A₂₀Noto
Label 1, a₁> A_TenAnd a₂> A₂₀Then label 2,
a₁<A_TenAnd a₂<A₂₀Then label 3, a₁<A_TenOr
Tied a₂> A₂₀Label 4, a₂₀<A₂<A₂₀When
Label 5. Labels 1 to 4 are smooth edge seeds
The label 5 is a roof edge. Label for Figure 40
The type of the demarcated boundary shape is shown. Step 3903 above
From step 3908 for all measurement points on the boundary
U In step 3909, the largest of the labels of all measurement points on the boundary is
Using the number M of labels with a large number and the boundary length K (Equation 10)
The ratio N of the largest number of labels is calculated by.

[0133]

[Equation 10]

If N is larger than the preset threshold value in step 3910, it is determined in step 3911 that the boundary shape is the smooth edge of the label type having the largest number, and if it is smaller, the roof edge is set in step 3912.

Next, a more detailed operation of the rounding radius estimating means 3721 will be described. FIG. 41 shows a flow of a method of estimating the round radius of the normal radius or the inverse radius when the rounding radius estimating means 3721 determines that the boundary is a smooth edge.

First, 3 of the point sequence measured in step 4101
Enter the dimensional coordinate value data. In step 4102, the boundary information of one area is input. In step 4103, the secondary difference value calculated in step 3905 and the linear parameter values a ₂ and b ₂ calculated in step 3907 are input. Then, step 410
In 4, the attention is paid to one pixel on the boundary. In procedure 4105, the second-order difference value of the measurement point sequence in the direction perpendicular to the boundary direction is substituted into (Equation 11) in order by the operator selected in procedure 3903 at the measurement point having the label determined in procedure 3910.

[0137]

[Equation 11]

In step 4106, if the value of t is -t ₀ <t <t ₀ with respect to the preset threshold value t ₀ (t ₀ > 0), then in step 4107, this measurement point is smooth edged. It is determined that it is the measurement point of the rounded part of and the process proceeds to the next measurement point. On the other hand, in step 4106, it is determined that the point where t ₀ <t or t <−t ₀ is the end point of the rounding, and
The substitution to 1) is stopped, and the number of measurement points substituted so far is stored in step 4108. A portion that is inward from the area boundary by the number of the measurement points substituted so far is called an internal area. The internal area indicates an area excluding the rounded portion from the entire area. In step 4109, the three-dimensional coordinate value data of the measurement points having the number of points determined in step 4104 in the direction perpendicular to the area boundary at the measurement points having the label determined in step 3910 are approximated to a circle by least squares. The radius thus obtained is the rounding radius. The above calculation is repeated for all measurement points on the boundary (procedures 4110 and 4111).

Next, the area type-based area boundary generation means 3
A more detailed operation at 722 will be described.

When the radius is a positive radius, it is not necessary for the parametric curved surface approximating means 3715 to approximate only the internal area to the parametric curved surface to create a special boundary. Only when the R is the inverse R, the area boundary is generated by the following procedure.

A detailed description will be given of a method of expanding the measurement data of the internal area excluding the rounded portion from the area so that the differential value becomes continuous when the radius is the inverse area to generate the area.

First, the internal region is roughly approximated to a spline curved surface that does not have multiple nodes at the end points. In order to extend the area on this spline curved surface by the number of points twice the number of measurement point sequences representing the rounded portion from the area boundary determined by the rounding radius estimation means 3721, input their xy coordinate values and Get the z coordinate value at a point. This operation expands the internal area. And the whole area including the expanded point sequence,
It is passed to the parametric curved surface approximating means 3715. By the above operation, the boundary type of each area is taken into consideration, and the area data in which the rounding operation can be easily corrected in the CAD can be created.

As described above, according to the present embodiment, the two colors of the surface color of the object to be measured and the reflectance with respect to the slit light are different from each other, and the reflectances are different from each other. An object to be measured 3704 and an object to be measured 370 in which a person roughly and thickly writes a curve for dividing a shape into areas and a curve used for parameter setting when each area is approximated to a curved surface.
Laser light source 3701 and a mirror 3702 as a slit light source for generating a laser slit light for irradiating 4 and a DUT 3704 are moved at a constant pitch in the X-axis direction.
An axis moving mechanism 3705, a camera 3706 for capturing the scattered light of the slit light by the DUT 3704, and the camera 37.
A / D converter 3707 for converting the output signal from 06 into a digitized image signal, an image memory 3708 for storing the image signal, and a central position of the slit scattered light from the image signal stored in the image memory 3708. The slit scattered light center position detecting means 3709 for detecting, the coordinate calculating means 3710 for calculating the three-dimensional coordinate value of the measured object from the slit scattered light center position, and the brightness value of the image signal are the measured object 37.
According to the curve indicating means B3711 for extracting the area dividing line based on the edge-likeness obtained from the measurement data from the pixel different from the surface color of No. 04 and its peripheral pixels, and the curve for area dividing indicated by the curve indicating means B3711 3 Area dividing means 3712 for dividing the dimensional coordinate value data into areas,
Boundary discriminating means 3720 for discriminating whether the boundary of the area is a roof edge, a smooth edge having a normal radius or a smooth edge having an inverse radius, a rounding radius estimating means 3721 for estimating a rounding radius of the regular radius and the inverse radius, and a regular radius portion Then, a rounded portion is deleted from the measurement data area to open an interval between the areas, and the measurement data of the internal area is expanded to generate an area so that the curvature is continuous in the inverse radius portion. And the parameters of the parametric curved surface in each area designated by the area-type-specific area boundary generation means 3722 are u and v, the parameters of the measurement points on the curve for the parameter setting designated by the curve designating means B3711. Parameter setting means 37 for giving a parameter to each measurement point so that the values of u or v are made uniform. 3, a node automatic determination means 3714 for automatically determining the direction and position of addition of a node and a control point in each of the divided areas, and each of the divided areas. In this case, the parametric curved surface approximating means 37 for sequentially approximating the three-dimensional coordinate value data of each region to the parametric curved surface using the nodes.
15, a data format conversion unit 3717 for converting the data compressed by the parametric curved surface approximation unit 3715 into the data format of the input destination system, and a scanner control unit 3716 for controlling the entire system. Input data to the three-dimensional CAD or computer graphics system 3719 in a data format that can efficiently reduce the amount of data while saving the measured shape and can easily change or correct the intended shape in the input destination system. be able to.

(Twelfth Embodiment) The twelfth embodiment of the present invention will be described below with reference to the drawings.

FIG. 42 is a block diagram of a three-dimensional shape input device according to the twelfth embodiment of the present invention. In FIG. 42,
Reference numeral 4201 denotes a laser light source for generating slit light for irradiating the object to be measured 4204. The object to be measured 4204 uses two colors having different surface colors of the object to be measured and reflectance with respect to the slit light, and further different from each other. A curve that divides the shape of the DUT 4204 in each color and a curve that is used for parameter setting when approximating each area to a curved surface are roughly and thickly written by a person. Reference numeral 4202 denotes a mirror for changing the optical path of slit light, 4203 denotes slit light for irradiating the DUT 4204, and 4205 denotes the DUT 4204 for X.
An X-axis moving mechanism that moves in a constant pitch in the axial direction, 420
6 is a camera for reading the scattered light of the slit light 4203, 4
Reference numeral 207 is an A / D converter that converts the output signal of the camera 4206 into a digitized image signal, 4208 is an image memory that stores the image signal, and 4209 is the slit scattered light center that calculates the center position of the slit scattered light from the image signal. The position detecting means 4210 is 3 from the center position of the slit scattered light.
Coordinate calculation means for calculating a dimensional coordinate value, reference numeral 4211 indicates two types of curves written on the DUT 4204 and 3 from the periphery thereof.
The curve extracting means B and 4212 for extracting the curve for dividing the area based on the edge-likeness obtained from the dimensional coordinate value data and the curve used for parameter setting are three-dimensional according to the curve for area division indicated by the curve indicating means 4211B. Area dividing means for dividing the coordinate value data into areas, 4220 is a boundary determining means for determining whether a boundary of the area is a roof edge, a smooth edge having a positive radius (convex curved surface) or a smooth edge having an inverse radius (concave curved surface), and 4221 denotes a boundary determining means. Rounding radius estimating means for estimating rounding radii of the normal radius and the reverse radius, 4
Reference numeral 213 indicates that, when the parameters of the parametric curved surface are u and v in the respective areas, each measurement is performed so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by the curve designating means B4211 are aligned. Parameter setting means for assigning a parameter to a point, reference numeral 4214 is an automatic node determination means for automatically determining the direction and position of addition of a node and a control point in each of the divided areas using the variance of the previous approximation error. , 4215 is parametric surface approximation means for sequentially approximating the three-dimensional coordinate value data of each area into a parametric surface using the nodes in each of the divided areas, and 4217 is parametric surface approximation means for each of the divided areas. Data for converting the data compressed by the 4215 into the data format of the input destination system Formatting conversion means 4216 scanner control unit for controlling the entire system, three-dimensional shape data 4218 obtained by the present apparatus, 4219 is a three-dimensional CAD system, or is (are) the same as the structure of FIG. 37.

37 is different from the structure shown in FIG. 37 in that the area type generating area 3722 shown in FIG. This is a point changed to the area boundary generation means 4222 for expanding the measurement data of the internal area to generate the area.

The operation of the three-dimensional shape input device configured as described above will be described. First, the DUT 42
The whole three-dimensional shape data of No. 04 is acquired, and the area dividing line is written on the DUT 4204 and the area dividing line is extracted from the surrounding area based on the edge likeness obtained from the measurement data by the curve indicating means B4211. The range of each area is determined, and in the same manner, a curve used for parameter setting when the surface is approximated to each area is extracted. The boundary discriminating means 4220 discriminates whether the boundary of the area is a roof edge, a smooth edge having a positive radius or a smooth edge having an inverse radius, and when the rounding radius estimating means 4221 determines that the boundary is a smooth edge, the regular radius is determined. Alternatively, the operation is the same as that of the eleventh embodiment until the round radius of the inverse radius is estimated.

Next, when the boundary is determined to be a smooth edge, the area boundary generating means 4222 regenerates the area by expanding the measurement data of the internal area so that the differential value of the internal area becomes continuous. Parameter setting means 4213, automatic node determination means 4214, parametric curved surface approximation means 421
5 compresses the shape information by individually approximating each area recreated by the area boundary generating means 4222 to a parametric curved surface. Data format conversion means 42
Reference numeral 17 converts the parameter values of a plurality of parametric curved surfaces representing the shape of the entire measured object approximated by the parametric curved surface approximating means 4215 into a data format for the CAD system of the input destination, and the format-converted three-dimensional shape data 4218 is 3D CAD system 4219
Entered in.

The more detailed operation of the area boundary generating means 4222 will be described below. Regardless of whether the radius is inverse radius or positive radius, the measurement data of the internal region is expanded so that the differential value is continuous to generate a region. First, the internal region is roughly approximated to a spline curved surface that does not have multiple nodes at the end points. Rounding radius estimation means 42 is applied to this spline curved surface.
To extend the area from the area boundary determined in step 21 by twice the number of measurement points representing the rounded portion, the x
-Enter the y coordinate value to obtain the z coordinate value at each point. This operation expands the internal area. Then, the entire area including the extended point sequence is set to parameter setting means, automatic node determination means 4214, and parametric curved surface approximation means 421.
Give to 5.

By the above-mentioned operation, the boundary type of each area is taken into consideration, and the area data in which the rounded portion is easily corrected can be created in the CAD. FIG. 43 shows an example in which area boundaries are generated. Areas 431, 432, and 433 are different areas. The intersection of the areas 431 and 432 represents the boundary when the reverse radius 434 is added, and the intersection of the areas 432 and 433 represents the boundary when the regular radius 435 is added.

As described above, according to the present embodiment, the two colors of the surface color of the object to be measured and the reflectance with respect to the slit light are different from each other and the reflectances are different from each other. An object to be measured 4204 and an object to be measured 420 in which a person roughly and thickly writes a curve for dividing a shape into areas and a curve used for parameter setting when approximating each area to a curved surface.
A laser light source 4201 and a mirror 4202 as a slit light source for generating a laser slit light for irradiating 4 and a DUT 4204 are moved at a constant pitch in the X-axis direction.
An axis moving mechanism 4205, a camera 4206 for capturing the scattered light of slit light by the DUT 4204, and a camera 42.
A / D converter 4207 for converting the output signal from 06 into a digitized image signal, an image memory 4208 for storing the image signal, and a central position of slit scattered light from the image signal stored in the image memory 4208. The slit scattered light center position detecting means 4209 for detecting, the coordinate calculating means 4210 for calculating the three-dimensional coordinate value of the measured object from the slit scattered light center position, and the brightness value of the image signal are the measured object 42.
According to the curve indicating means B4211 for extracting the area dividing line based on the edge likeness obtained from the measurement data from the pixel different from the surface color of 04 and the peripheral pixels, and the curve for area dividing indicated by the curve indicating means B4211 3 Region dividing means 4212 for dividing the dimensional coordinate value data into regions,
Boundary discriminating means 4220 for discriminating whether the boundary of the area is a roof edge, a smooth edge of a normal radius or a smooth edge of a reverse radius, a rounding radius estimating means 4221 for estimating a rounding radius of a normal radius and a reverse radius, and a region boundary are In the case of a smooth edge, an area boundary generation unit 4222 that expands the measurement data of the internal area so as to make the differential value continuous from the internal area to generate an area, and each area specified by the area boundary generation unit 4222. If the parameters of the parametric surface are u and v, the curve designating means B42
Parameter setting means 4 for assigning a parameter to each measurement point so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by 11 are aligned.
213, a nodal point automatic determination means 4214 for automatically determining the direction and position of addition of a nodal point and a control point in each of the divided areas, and each of the divided areas. , The parametric curved surface approximating means 4215 for sequentially approximating the three-dimensional coordinate value data of each area to the parametric curved surface using the above-mentioned nodes, and the parametric curved surface approximating means 42 for each of the divided areas.
Data format conversion means 4 for converting the data compressed by 15 into the data format of the input destination system
217 and scanner control means 4216 for controlling the entire system
By providing the, the amount of data can be efficiently reduced while saving the measured shape, and it is easy to change or correct the intended shape in the input destination system. Is a data format that can be easily performed using CAD internal commands
D or computer graphics system 4219.

(Embodiment 13) Hereinafter, a thirteenth embodiment of the present invention will be described with reference to the drawings.

FIG. 44 is a block diagram of the three-dimensional shape input device in the thirteenth embodiment of the present invention. In FIG. 44,
Reference numeral 4401 is a laser light source for generating slit light for irradiating the object to be measured 4404. The object to be measured 4404 uses two colors having different surface colors of the object to be measured and the reflectance with respect to the slit light and further different in reflectance from each other. A curve that divides the shape of the DUT 4404 in each color and a curve that is used for parameter setting when approximating each area to a curved surface are roughly and thickly written by a person. Reference numeral 4402 denotes a mirror for changing the optical path of the slit light, 4403 denotes slit light for irradiating the DUT 4404, and 4405 denotes the DUT 4404 for X.
X-axis moving mechanism 440 that moves in a constant pitch in the axial direction
6 is a camera for reading the scattered light of the slit light 4403, 4
Reference numeral 407 is an A / D converter for converting the output signal of the camera 4406 into a digitized image signal, 4408 is an image memory for storing the image signal, 4409 is a slit scattered light center for calculating the central position of the slit scattered light by the image signal. The position detecting means, 4410, is 3 from the center position of the slit scattered light.
Coordinate calculation means for calculating a dimensional coordinate value, reference numeral 4411 indicates two types of curves written on the object to be measured 4404 and their surroundings.
The curve extracting means B and 4412 for extracting the curve for dividing the area based on the edge-likeness obtained from the dimensional coordinate value data and the curve used for parameter setting are three-dimensional according to the curve for area division indicated by the curve indicating means 4411B. Area dividing means for dividing the coordinate value data into areas, 4420 is a boundary determining means for determining whether a boundary of the area is a roof edge, a smooth edge having a positive radius (convex curved surface) or a smooth edge having an inverse radius (concave curved surface), 4421 is Rounding radius estimating means for estimating rounding radii of the normal radius and the reverse radius, 4
If the region boundary is a smooth edge, 422 is a region boundary generation means for expanding the measurement data of the internal region to generate a region so that the differential value is continuous regardless of whether the radius is a normal radius or an inverse radius. If the parameters of the parametric surface are u and v, the curve designating means B441
Parameter setting means for assigning a parameter to each measurement point so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by 1 are aligned
Reference numeral 414 denotes a nodal point automatic determination means for automatically determining the direction and position of addition of a nodal point and a control point in each of the divided areas, and 4415 denotes each of the divided areas. Parametric curved surface approximating means for sequentially approximating the three-dimensional coordinate value data of each area to a parametric curved surface using the nodes, 4417 indicates parametric curved surface approximating means 4415 for each of the divided areas.
Data format conversion means for converting the data compressed by the data format of the input destination system, 44
Reference numeral 16 is a scanner control means for controlling the entire system, 4418 is three-dimensional shape data obtained by this apparatus, 4419 is a three-dimensional CAD system, and the above is the same as the configuration of FIG.

The difference from the configuration of FIG. 42 is that a region combining unit 4223 is newly provided for performing a rounding deformation operation based on the rounding radius estimated by the rounding radius estimating unit 4421 to join regions.

The operation of the three-dimensional shape input device configured as described above will be described below.

First, the three-dimensional shape data of the entire object to be measured 4404 is acquired, and the object to be measured 4404 is measured based on the edge likelihood obtained from the measurement data by the curve designating means B4411.
The area division line is extracted from the area division instruction line and its surroundings, the range of each area is determined, and similarly, a curve used for parameter setting when approximating a curved surface of each area is extracted. The boundary discriminating means 4420 discriminates whether the boundary of the area is a roof edge, a smooth edge having a normal radius or a smooth edge having an inverse radius, and a rounding radius estimating means 4421.
If it is determined that the boundary is a smooth edge, the rounding radius of the normal radius or the inverse radius is estimated, and if the boundary is determined to be a smooth edge, the differential value of the internal region is continuous. The measurement data of the internal region is expanded so as to generate the region again, and the parameter setting unit 4413, the automatic node determination unit 4414, and the parametric curved surface approximation unit 4415 are used as the region boundary generation unit 4.
The operation is the same as that of the twelfth embodiment until the shape information is compressed by individually approximating each area recreated by 422 to a parametric curved surface.

Next, the area combining means 4423 connects the areas by creating the radius of the area boundary with the maximum value of the rounding radius estimated by the rounding radius estimating means 4421. The data format converting means 4417 inputs the parameter values of a plurality of parametric curved surfaces representing the shape of the entire object to be measured as C at the input destination.
The three-dimensional shape data 4418 converted into the data format for the AD system and converted in the format is the three-dimensional C
It is input to the AD system 4419.

The more detailed operation of the area combination means 4423 will be described below. FIG. 45 shows a rounding radius estimating means 442.
2 shows a flow of a method of creating each radius based on the rounding radius estimated by 1 and combining the regions.

First, the parameter values of the two surfaces to be combined in step 4501 and the radius type and the rounding radius estimated by the rounding radius estimating means 4421 are input. In step 4502, the number of divisions of the ridgeline to be rounded is set.
The greater the number of divisions, the more accurate are are generated. Parameters from one end of the ridgeline to the other are 0.0 to 1.
When the value is 0, the parameter value s is first set to 0.0. In addition, a point on the edge where the parameter value s is 0.0 is P
Set to 0. In step 4503, it is determined whether the parameter value s is larger than 1.0. If so, go to step 4511. If it is smaller, the procedure proceeds to step 4504. Step 4504
Then, on the two parametric surfaces, the initial point P11 (parameter (u11, v1
1)) and P12 (parameter (u12, v12)) are set. In step 4505, the tangent planes T1 and T2 at P11 and P12 are calculated. The tangent plane is created on the side where the radius is created. In step 4506, a plane T0 passing through the point P0 and perpendicular to T1 and T2 is obtained. A sphere having a radius of the rounding radius estimated by the rounding radius estimating means 4421 in step 4507 is placed in contact with T1 and T2, and the center of the sphere is on T0, and the contact point P11 ′ between the sphere and T1 and T2 is Calculate P12 'respectively. FIG. 46 shows a state in which the sphere is placed. In step 4508, P11, P11 ', and P12
And P12 ′ are the same point, the point is the contact point between the sphere and the two parametric curved surfaces, and the procedure proceeds to step 4509. If they are not the same point, in step 4516, P11 and P1
2 is stored, the division point of the ridge line parameter s is moved to the next division point, and the process returns to step 4503. Point P11'P1 in step 4509
Next starting point P2 on two parametric surfaces from 2 '
1 (parameter (u21, v21)) and P22 (parameter (u22, v22)) are obtained. The inner products of (Equation 12) and (Equation 13) and (Equation 12) and (Equation 14) are respectively taken to obtain Δu1 and Δv1 from the binary linear simultaneous equations. Then, u21 and v21 are obtained by (Equation 15) and (Equation 16).

[0160]

[Equation 12]

[0161]

[Equation 13]

[0162]

[Equation 14]

[0163]

[Equation 15]

[0164]

[Equation 16]

The same applies to u22 and v22. Step 451
At 0, the values of u21, v21 are changed to u11, v11, u22,
The value of v22 is substituted for u12 and v12, and the procedure returns to step 4505. In step 4511, a curve passing through the point sequence stored as P11 in step 4508 is generated. A similar curve is generated for P12. Step 4512 is step 4
A new vertex is created at the point where the curve obtained in 511 intersects the ridge of the surface of the original object to be measured. In Step 4513, Step 4
A plurality of ridge lines connecting the new vertices obtained at 512 are created. In step 4514, the original intersection line of the curved surface is deleted. Step 4
In 515, an arc ridge connecting the ridges created in step 4513 is generated. With the above operation, the boundary of the rounded portion is generated. This state is shown in FIG. Finally, the rounded part is interpolated from the boundary. There are various rounding methods, and the rounding method described in the present embodiment does not limit the present invention.

As described above, according to the present embodiment, the two colors of the surface color of the object to be measured and the reflectance with respect to the slit light are different from each other, and the reflectances are different from each other. An object to be measured 4404 and an object to be measured 440 in which a person roughly and thickly writes a curve for dividing a shape into areas and a curve used for parameter setting when approximating each area to a curved surface.
Laser light source 4401 and a mirror 4402 as a slit light source for generating a laser slit light for irradiating 4 and a DUT 4404 are moved at a constant pitch in the X-axis direction.
An axis moving mechanism 4405, a camera 4406 for capturing the scattered light of the slit light by the DUT 4404, and the camera 44.
A / D converter 4407 for converting the output signal from 06 into a digitized image signal, an image memory 4408 for storing the image signal, and a central position of the slit scattered light from the image signal stored in the image memory 4408. The slit scattered light center position detecting means 4409 for detecting, the coordinate calculating means 4410 for calculating the three-dimensional coordinate value of the measured object from the slit scattered light center position, and the luminance value of the image signal are the measured object 44.
According to the curve indicating means B4411 for extracting the area dividing line based on the edge likeness obtained from the measurement data from the pixel different from the surface color of 04 and the peripheral pixels, and the curve for area dividing indicated by the curve indicating means B4411 3 Region dividing means 4412 for dividing the dimensional coordinate value data into regions,
Boundary discriminating means 4420 for discriminating whether the boundary of the area is a roof edge, a smooth edge of a normal radius or a smooth edge of a reverse radius, a rounding radius estimating means 4421 for estimating a rounding radius of a normal radius and a reverse radius, and a region boundary are When the edge is a smooth edge, an area boundary generation unit 4422 that expands the measurement data of the internal area to generate an area so that the differential value is continuous from the internal area, and each area specified by the area boundary generation unit 4422. If the parameters of the parametric surface are u and v, the curve designating means B44
Parameter setting means 4 for assigning a parameter to each measurement point so that the values of the parameters u or v of the measurement points on the curve for parameter setting designated by 11 are aligned.
413, a node automatic determination means 4414 for automatically determining the direction and position of addition of a node and a control point in each of the divided areas using the variance of the previous approximation error, and each of the divided areas. In this case, the three-dimensional coordinate value data of each area is sequentially created by using the nodes, and each radius is created based on the rounding radius estimated by the rounding radius estimating means 4421 and the parametric curved surface approximating means 4415 that approximates the parametric curved surface. Area joining means 44 for joining areas
23, the data format conversion means 4417 for converting the data compressed by the parametric curved surface approximation means 4415 into the data format of the input destination system, and the scanner control means 4416 for controlling the entire system to save the measured shape. It is easy to understand the shape on the CAD by creating a rounded part that is difficult to create accurately by clay mock-up etc. It is easy to change and modify the shape, and it is possible to easily modify the rounded portion, which is frequently modified, by using the internal command of CAD.

[0167]

As described above, according to the present invention, firstly, the slit light source for generating the slit light for irradiating the measured object, the moving mechanism for moving the measured object, and the scattering of the slit light from the measured object. A camera for capturing light, an A / D converter for converting a luminance signal from the camera into a digitized image signal, an image memory for storing the image signal, and a slit from the image signal stored in the image memory Slit scattered light center position detecting means for detecting the center position of scattered light, coordinate calculating means for calculating three-dimensional coordinate values from the slit scattered light center position, curve indicating means for indicating an edge on the object to be measured, and parametric If the parameters of the curved surface are u and v, u of the parameter values of the measurement points on the continuous edges are
Alternatively, parameter setting means for giving a parameter to each measurement point so as to make the values of v uniform and three-dimensional coordinate value data representing the entire shape of the object to be measured increase the number of nodes until a function less than a specified approximation error is obtained. When the three-dimensional coordinate value data is approximated to a parametric surface by repeating movement and increase in the number of control points, the direction of defining the nodes is defined as u, v as a method of automatically determining the increasing direction and the position of the nodes and control points. Then, for each set U _i of measurement point sequences having the same value of the parameter u and each set V _{j of} measurement point sequences having the same value of the parameter v,
Variance Uvar _i (i = 1, ..., M), Vva of the difference in depth coordinate value between the previous approximation and the three-dimensional coordinate value data
r _j (j = 1, ..., N) is calculated, and the variance U is calculated.
The averages U and V of var _i and Vvar _j are obtained, and their magnitudes are compared. If U is large, it is decided to increase the nodes and control points in the v direction, and if V is large, it is decided to increase the nodes and control points. For example, Uvar _i (i = 1, ...,
m), a set U ₀ of measurement point sequences having the largest variance value is obtained, the sum of squares of the error of the point sequence U ₀ is calculated for each node, and the center between the nodes having the largest sum of squares is calculated. If the number of new nodes is increased and the rate of approximation error reduction is low even if the number of control points is increased, the surface of the object to be measured is divided into multiple parts in the direction where the fluctuation of the error is large, and the parametric function approximation is performed as different curved surfaces. If the rate of decrease of the approximation error is considerably low even if the number of control points is increased, data compression means for creating a polygon patch using the feature points of the measured three-dimensional coordinate value data to construct a surface, and By providing a data format conversion means for converting the data compressed by the data compression means into the data format of the input destination system, the measured shape can be saved in an efficient amount of data. You can reduce and smooth the data.

Secondly, a slit light source for generating slit light for irradiating the object to be measured, a moving mechanism for moving the object to be measured, a camera for capturing scattered light of the slit light from the object to be measured, and the camera A / D converter for converting the luminance signal of the above into a digitized image signal, an image memory for storing the image signal, and slit scattering for detecting the central position of slit scattered light from the image signal stored in the image memory If the light center position detecting means, the coordinate calculating means for calculating the three-dimensional coordinate value from the slit scattered light center position, the curve indicating means for indicating the edge on the object to be measured, and the parameter of the parametric curve are u, they are continuous. Parameter setting means for assigning a parameter to each measurement point so that the values of the parameter u of the measurement point on the edge are aligned, and a cubic representing the overall shape of the measured object. Increasing positions of nodes and control points when approximating three-dimensional coordinate value data to a plurality of parametric curves until the coordinate value data is repeated by increasing and moving the number of control points and increasing the number of control points until a function below a specified approximation error is obtained. As a method of automatically determining, the sum of squares of the approximation error of the point sequence U that approximates one parametric curve is calculated for each node, and a new node is added to the center between the nodes with the largest sum of squares. If the rate of decrease in the approximation error is low even if the number of control points is increased, the curve is divided into multiple curves and the parametric curve approximation is performed as different curves.
Next, by providing data compression means for interpolating these plural parametric curves to create a curved surface, and data format conversion means for converting the data compressed by the data compression means into the data format of the input destination system, It is possible to reduce the shape of the undulation that normally occurs when performing curved surface approximation, and the number of control points and the positions of nodes to obtain a more efficient approximation that was difficult to predict directly from the shape of the measured object. It is possible to solve the above problem and approximate three-dimensional coordinate value data to a parametric curved surface with a data amount as small as possible.

Thirdly, in addition to the above-mentioned first and second constituent elements, a curve designating means for determining a curve for dividing the object to be measured into a plurality of areas, and the curve to divide the object to be measured into a plurality of areas. And a data compression means for approximating the three-dimensional coordinate value data to a parametric function for each area. However, the curve designating means used at the time of parameter setting and area division specifies the curve on the object to be measured using a color whose reflectance to the slit light is different from the surface color of the object to be measured, and the brightness value in the image signal. Is extracted as the curve, or the reflectance to the slit light is specified by using a color different from the surface color of the DUT to specify the rough curve on the DUT and the brightness value in the image signal. Is extracted based on the edge-likeliness obtained using the three-dimensional coordinate value data from the different portion and its peripheral portion, or the obtained curve is regarded as the edge portion of the object to be measured, and human intervention is required. Without the need to automatically obtain the edge using the three-dimensional coordinate value data, or the camera reads the slit scattered light and does not project the slit light. A memory switching unit for capturing an image of the whole and switching the output depending on whether the camera captures the slit light or the luminance information of the entire object to be measured which is not projecting the slit light. A first image memory for storing the image signal of the slit scattered light by switching, and a second image memory for storing the brightness image of the entire measured object not projecting the slit light by switching the memory switching means. Is provided separately, the curve to be obtained is regarded as the edge portion of the object to be measured, and the edge is automatically obtained using the brightness image without human intervention, or a color different from the surface color of the object to be measured is obtained. The above curve is designated on the object to be measured by using the camera to read the slit scattered light and to image the entire object to be measured in the state where the slit light is not projected. Camera, the memory switching means for switching the output depending on whether the camera images the slit light or the luminance information of the entire object to be measured which does not project the slit light, and the slit by switching the memory switching means. A first image memory for storing an image signal of scattered light and a second image memory for storing a brightness image of the entire measured object not projecting slit light by switching the memory switching means are separately provided. , The curve was extracted from the luminance image using the color difference or measured 3
The dimensional coordinate value data is displayed on a computer, and a human uses the mouse to indicate a curve for the displayed three-dimensional coordinate value data. By providing the above configuration, it is possible to easily specify a curve that divides the entire shape of the DUT into a plurality of areas and a curve that is used by the parameter setting means when efficiently approximating each area to a parametric curved surface.

Fourthly, in addition to the above-mentioned third component, boundary discrimination for discriminating whether the boundary portion of the area divided by the area dividing means is the roof edge, the smooth edge of the convex curved surface or the smooth edge of the concave curved surface. And a rounding radius estimating means for estimating rounding radii of the convex curved surface and the concave curved surface, and an area boundary generating means for regenerating a special area boundary in the convex curved surface and the concave curved surface. Further, in addition to the above components, by changing the shape in the input destination system by providing a configuration with a region joining unit that joins regions by performing a rounding deformation operation based on the rounding radius estimated by the rounding radius estimating unit, It is possible to realize an excellent three-dimensional shape input device capable of inputting into a three-dimensional CAD system or a computer graphics system in a data format that can be easily corrected.

[Brief description of drawings]

FIG. 1 is a block connection diagram of a three-dimensional shape input device according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an instruction of a curve used for parameter setting in the same three-dimensional shape input device.

FIG. 3 (a) is a diagram showing a case where one curved line in the same three-dimensional shape input device does not have one point per frame. (B) A designated curve in the same three-dimensional shape input device is converted into a plurality of curves. Diagram showing the case of division

FIG. 4 (a) is a diagram showing a case where both end points of a curve in the same three-dimensional shape input device are not placed in the first frame and the final frame. (B) Both end points in the same three-dimensional shape input device are first frame And the figure that shows the case of reconstructing a curve that is not on the final frame

FIG. 5 is a conceptual diagram for setting a value of a parameter v corresponding to a position of a designated curve in the same three-dimensional shape input device.

FIG. 6 is a conceptual diagram of setting a value of a parameter v of a data point on one slit in the same three-dimensional shape input device.

FIG. 7 is a procedure flow chart of a method for approximating the data in the three-dimensional shape input device to a parametric curved surface with a specified accuracy or less.

FIG. 8 is a block connection diagram of a three-dimensional shape input device according to a second embodiment of the present invention.

FIG. 9 is a flowchart of a termination condition of successive approximation in the same three-dimensional shape input device.

FIG. 10 (a) x, y in the same three-dimensional shape input device
Figure of the original data when showing the difference in the appearance of the residual of the coordinate values and the residual of the z coordinate values. (B) The residual of the x, y coordinate values and the z coordinate value of the same three-dimensional shape input device. X, y when showing the difference in how residuals appear
Figure of approximation when residual of coordinate values is large (c) z coordinate value when showing difference in appearance of residual of x and y coordinate values and residual of z coordinate value in the same three-dimensional shape input device Figure of approximation when residual of x is large and residual of x and y coordinate values is small

FIG. 11 is a relational diagram of parameters of measurement data points and reference grid parameters in the same three-dimensional shape input device.

FIG. 12 (a) is a diagram showing a case where the residuals in the neighborhood region of the same three-dimensional shape input device have different signs and have the same absolute values. (B) Of the residuals in the neighborhood region of the same three-dimensional shape input device Diagram showing the case where the sign and the absolute value are the same

FIG. 13 is a conceptual diagram of a point sequence for calculating variance in the same three-dimensional shape input device.

FIG. 14 is a conceptual diagram of a direction in which nodes are added in the same three-dimensional shape input device.

FIG. 15 is a conceptual diagram of positions at which nodes are added in the same three-dimensional shape input device.

FIG. 16 is a block connection diagram of a three-dimensional shape input device according to a third embodiment of the present invention.

FIG. 17 is a procedure flow chart of a method of dividing and approximating data into a parametric curved surface with specified accuracy or less on the same three-dimensional shape input device.

FIG. 18 is a procedure flow chart of a feature point extraction method in the same three-dimensional shape input device.

FIG. 19 is a conceptual diagram of feature point extraction in the same three-dimensional shape input device.

FIG. 20 is a procedure flow chart of a method for generating a polygon patch from a sequence of points on two slits in the same three-dimensional shape input device.

FIG. 21 is a conceptual diagram showing generation of a triangular patch in the same three-dimensional shape input device.

FIG. 22 is a block connection diagram of a three-dimensional shape input device according to a fourth embodiment of the present invention.

FIG. 23 is a procedure flow chart of a method for approximating the data of one slit in the three-dimensional shape input device to a parametric curve with a specified accuracy or less.

FIG. 24 is a block connection diagram of a three-dimensional shape input device according to a fifth embodiment of the present invention.

FIG. 25 is a block connection diagram of a three-dimensional shape input device according to a sixth embodiment of the present invention.

FIG. 26 is a procedure flow chart of a method of extracting a region dividing line from a peripheral region of a designated curve in the same three-dimensional shape input device.

FIG. 27 is a diagram of a Kirsch operator in the same three-dimensional shape input device.

FIG. 28 is a conceptual diagram of erasability in thinning the same three-dimensional shape input device.

FIG. 29 is a block connection diagram of a three-dimensional shape input device according to a seventh embodiment of the present invention.

FIG. 30 is a procedure flow chart of an automatic region segmentation method by three-dimensional data in the same three-dimensional shape input device.

FIG. 31 is a conceptual diagram of label division of divided areas in the same three-dimensional shape input device.

FIG. 32 is a block connection diagram of a three-dimensional shape input device according to an eighth embodiment of the present invention.

FIG. 33 is a procedure flow chart of an automatic area dividing method using luminance data in the same three-dimensional shape input device.

FIG. 34 is a block connection diagram of a three-dimensional shape input device according to a ninth embodiment of the present invention.

FIG. 35 is a procedure flow chart of an automatic curve extraction method by a color image in the same three-dimensional shape input device.

FIG. 36 is a block connection diagram of the three-dimensional shape input device according to the tenth embodiment of the present invention.

FIG. 37 is a block connection diagram of a three-dimensional shape input device according to an eleventh embodiment of the present invention.

FIG. 38 is a conceptual diagram of boundary type for each radius in the same three-dimensional shape input device.

FIG. 39 is a procedure flow chart of the boundary discriminating means in the same three-dimensional shape input device.

FIG. 40 is a conceptual diagram of labels attached to area boundaries in the same three-dimensional shape input device.

FIG. 41 is a procedure flow chart of a rounding radius estimating means in the same three-dimensional shape input device.

FIG. 42 is a block connection diagram of the three-dimensional shape input device according to the twelfth embodiment of the present invention.

FIG. 43 is a conceptual diagram of region boundary generation by the region boundary generation means in the same three-dimensional shape input device.

FIG. 44 is a block connection diagram of a three-dimensional shape input device according to a thirteenth embodiment of the present invention.

FIG. 45 is a procedure flow chart of a region combination means in the same three-dimensional shape input device.

FIG. 46 is a conceptual diagram of processing of a region joining unit in the same three-dimensional shape input device.

FIG. 47 is a conceptual diagram of region combination by the region combination means in the same three-dimensional shape input device.

FIG. 48 is a block connection diagram of a conventional three-dimensional shape input device.

[Explanation of symbols]

 101 laser light source 102 vibrating mirror 103 slit light 104 object to be measured 105 X-axis moving mechanism 106 camera 107 A / D converter 108 image memory 109 slit scattered light center position detecting means 110 coordinate calculating means 111 coordinate calculating means 112 parameter setting means 113 Parametric curved surface approximation means 114 Data format conversion means 115 Scanner control means 116 Three-dimensional shape data 117 obtained by this device 117 Three-dimensional CAD system

Claims (21)

[Claims] 1. A slit light source for generating slit light for irradiating an object to be measured, a moving mechanism for moving the object to be measured, a camera for capturing scattered light of the slit light from the object to be measured, and the camera. A / D converter for converting the luminance signal from the image signal into a digitized image signal, an image memory for storing the image signal, and a slit for detecting the center position of the slit scattered light from the image signal stored in the image memory a scattered light center position detection means, and coordinate calculation means for calculating three-dimensional coordinates from the center of the slit scattered light, before
Parametric curved surface of 3D coordinate data of the object to be measured
A three-dimensional shape input device comprising: a data compression unit that is similar to the above, and a data format conversion unit that converts the data compressed by the data compression unit into the data format of the input destination system. 2. The three-dimensional shape input device according to claim 1, further comprising slit light scanning means for scanning the slit light on the measured object instead of the moving mechanism for moving the measured object. 3. A curve designating means for designating a curve on the object to be measured, and u and v as parameters of the parametric curved surface. The parameter values u or v of the measurement points on the continuous curve are aligned. The three-dimensional shape input device according to claim 1 or 2 , further comprising: parameter setting means for giving a parameter to each measurement point. 4. A bidirectional node when the data compressing means approximates three-dimensional coordinate value data to a parametric curved surface.
The three-dimensional shape input device according to claim 1, 2 or 3 , wherein the increase and the movement of the control points are repeated until a curved surface having a specified approximation error or less is obtained. 5. The data compressing means automatically determines the increasing directions and positions of the nodes and the control points when the three-dimensional coordinate value data is repeatedly increased and moved for the nodes and the control points to approximate the parametric curved surface. As a method, if the directions defining the nodes are u and v, for each set U _i of measurement point sequences in which the value of the parameter u is equal and V _j of measurement point sequences in which the value of the parameter v is equal, Variance Uvar _i (i = 1, ...) Of difference in depth coordinate value from the three-dimensional coordinate value data
.., m), Vvar _j (j = 1, ..., N) are calculated, the averages U and V of the variances Uvar _i and Vvar _j are obtained, and the magnitudes are compared. If U is large, V direction, V
Is larger, it is decided to increase the number of nodes and control points in the u direction. For example, Uvar _i
At (i = 1, ..., M), a set U ₀ of measurement point sequences having the largest variance value is obtained, and the sum of squares of the error of the point sequence U ₀ is calculated for each node to obtain the most square. claims 1 to increase the new nodal point at the center between the nodes the value of the sum is greater
The three-dimensional shape input device according to item 4 . 6. The data compression means comprises bidirectional nodes and control.
When the approximation error decreases until the approximation error becomes less than or equal to the specified value by repeating the increase and the movement of the points, and if the rate of decrease of the approximation error is low even if the control points are increased, the error of the curved surface is changed. Is divided into multiple parts in the direction where the
This operation is repeated until a curved surface with the specified approximation accuracy is obtained.
The three-dimensional shape input device according to claim 4 or 5 . 7. A bidirectional node for the data compression means to perform curved surface approximation with an approximation error of a specified value or less for each divided area.
When the control points are repeatedly increased and moved , the approximation error decreases at a low rate even if the control points are increased, or the approximation error has a limit and the specified approximation accuracy cannot be achieved. 7. The three-dimensional shape input device according to claim 6 , wherein a polygonal patch is created by using to construct a surface. 8. A bidirectional node for the data compression means to perform curved surface approximation with an approximation error of a specified value or less for each divided area.
When the control points are repeatedly increased and moved , the approximation error decreases at a low rate even if the control points are increased, or the approximation error has a limit and the specified approximation accuracy cannot be achieved. A contract for extracting a feature point from the image and creating a polygonal patch using the feature point to construct a surface.
The three-dimensional shape input device according to claim 6 . 9. The three-dimensional according to claim 1, wherein the data compression means approximates the three-dimensional coordinate value data to a plurality of parametric curves and then creates a curved surface for interpolating the plurality of parametric curves. Shape input device. 10. The data compression means, when approximating the three-dimensional coordinate value data to a plurality of parametric curves, repeats increasing and moving of nodes and control points of the parametric curve until a curve having a specified approximation error or less is obtained. Claim 9
3-dimensional shape input device of the mounting. 11. The data compressing means increases and shifts nodes and control points in order to perform curve approximation with an approximation error of a specified value or less.
If the rate of decrease of the approximation error is low even when the control points are increased when repeating the motion , divide the curve into multiple curves and perform parametric curve approximation as different curves until a curve with the specified approximation accuracy is obtained. Repeat claim 10
The three-dimensional shape input device described . 12. A curve indicating means for determining a curve for dividing an object to be measured into a plurality of areas, and an area dividing means for dividing the object to be measured into a plurality of areas by the curve are provided, and the data compressing means is provided for each area. The three-dimensional coordinate value data is approximated to a parametric function according to claim 1, 2, 3, 4, 5, 6, 7,
The three-dimensional shape input device according to 8, 9, 10 or 11 . 13. An object to be measured whose curve is designated by using a color whose reflectance to slit light is different from the surface color of the object to be measured, wherein the curve indicating means has a different brightness value in the image signal. claim extracted as the curve 3 or 1
2 three-dimensional shape input device according. 14. An object to be measured having a rough curve is provided by using a color whose reflectance to slit light is different from the surface color of the object to be measured, and the curve indicating means has a brightness value in the image signal. From the different part and its peripheral part, the above 3
The three-dimensional shape input device according to claim 3 or 12 , wherein the curve is extracted based on the edge-likeness obtained using the three-dimensional coordinate value data. 15. Curve indicating means, the curve determining the regarded as the edge portion of the object to be measured, using automatically the 3-dimensional coordinate value data without intervening manual seek edge claim 3 or 12. The three-dimensional shape input device according to item 12 . 16. A camera reads the slit scattered light and picks up the entire object to be measured in a state where the slit light is not projected, and the camera picks up the slit light or does not project the slit light. A memory switching unit that switches the output depending on whether to capture the luminance information of the entire object to be measured, a first image memory that stores the image signal of the slit scattered light by switching the memory switching unit, and a switching of the memory switching unit. And a second image memory for storing a brightness image of the entire object to be measured which is not projecting the slit light, and the curve indicating means considers the curve to be an edge portion of the object to be measured, and The three-dimensional shape input device according to claim 3 or 12 , wherein an edge is automatically obtained using the luminance image without intervening. 17. The object to be measured has the curve designated by using a color different from the surface color of the object to be measured, and the camera reads the slit scattered light and does not project the slit light. A color camera for imaging the entire measurement object, a memory switching unit for switching the output depending on whether the camera captures the slit light or the brightness information of the entire measurement object that does not project the slit light, A first image memory that stores the image signal of the slit scattered light by switching the memory switching unit, and a second image memory that stores the entire brightness image of the measured object that does not project the slit light by switching the memory switching unit. an image memory provided, the three-dimensional shape input of claim 3 or 12, wherein <br/> curve indicating means for extracting the curve using the difference in color from the luminance image Location. 18. The curve designating means displays the measured three-dimensional coordinate value data on a computer, and the displayed 3
The three-dimensional shape input device according to claim 3 or 12, wherein a human uses a mouse to indicate a curve for the three-dimensional coordinate value data. 19. A boundary discriminating means for discriminating between a roof edge, a smooth edge of a convex curved surface, or a smooth edge of a concave curved surface, and a rounding radius of a convex curved surface and a concave curved surface. Rounding radius estimating means for estimating, and a convex curved surface, the rounded portion is deleted from the three-dimensional coordinate value data of the area to provide an interval between the areas, and a concave curved surface is an area excluding the rounded portion so that the differential values are continuous. claim provided with the 3-dimensional coordinate values rounded typing region boundary generating means data expanded by the regenerate area 12,1
The three-dimensional shape input device according to 3, 14, 15, 16, 17 or 18 . 20. Boundary discriminating means for discriminating whether a boundary of the area divided by the area dividing means is a roof edge, a smooth edge of a convex curved surface or a smooth edge of a concave curved surface, and a rounding radius of a convex curved surface and a concave curved surface is estimated. Rounding radius estimating means and a region boundary generating means for expanding the three-dimensional coordinate value data of the area excluding the rounded part so that the differential value is continuous at the smooth edge and newly generating the area. Claims 12, 13, 14, 15, 16, 17 or
18. The three-dimensional shape input device according to item 18 . 21. The three-dimensional shape input device according to claim 19 , further comprising area joining means for joining the areas by performing a rounding deformation operation based on the rounding radius estimated by the rounding radius estimating means.