TWI416360B - Computer system and method for fitting characteristic elements - Google Patents

Abstract

The present invention provides a method for fitting characteristic elements. The method includes: constructing triangles by using point cloud data of a measured object; selecting points to be fit the characteristic elements from the triangles; obtaining uppermost layer points of the selected points; extracting boundary points from the uppermost layer points; fitting the boundary points to a characteristic element; obtaining parameters of the characteristic element, and drawing a figure according to the characteristic element and the parameters. A related computer system for fitting characteristic elements is also disclosed.

Description

Feature element fitting method and computer system thereof

The invention relates to a feature element fitting method and a computer system thereof.

In the development and manufacturing process of 3D products, error analysis has become an important part of product quality verification. The traditional approach is to use the point cloud acquisition device to obtain the point cloud data of the workpiece (ie, a collection of points consisting of multiple three-dimensional discrete points), and then log the point cloud data to the computer, and execute the corresponding software to process the point cloud data, which will be discrete. The points are fitted and created as feature elements such as lines, faces, circles, cylinders or spheres. The actual artifacts are then aligned with the feature elements in the theoretical design file to determine if there is an error between the actual artifact and the image in the theoretical design image.

At present, there are many methods for fitting feature elements on the market, for example, using the least squares method to fit feature elements, but these methods cannot quickly extract the uppermost point of the point cloud and the effective boundary point in the three-dimensional coordinate system.

In view of the above, it is necessary to provide a feature element fitting method and a computer system thereof, which can quickly extract the uppermost point of the point cloud and the effective boundary point in the three-dimensional coordinate system, thereby completing the fitting of the feature elements.

A feature element fitting method, the method comprising: establishing a triangular mesh surface according to the received point cloud data; selecting a point cloud from the triangular mesh surface to be fitted with the feature element; obtaining from the selected point cloud An uppermost point of the screen; extracting a boundary point in the uppermost point; fitting the boundary point in the uppermost point to a feature element by using an iterative method; acquiring the fitted feature A parameter of the element; and a feature element is drawn according to the parameter of the feature element and the feature element that is fitted.

A computer system for fitting a feature element, the computer system comprising: a triangular meshing module for establishing a point cloud into a triangular mesh surface; and an extraction module for selecting a fitting from the triangular mesh surface a point cloud of the feature element, obtaining an uppermost point from the selected point cloud, and extracting a boundary point in the uppermost point; a fitting module for using an iterative method to select a boundary point in the uppermost point Fitting into a feature element and acquiring parameters of the fitted feature element; and creating a module for drawing the feature element according to the parameter of the feature element and the fitted feature element.

Compared with the prior art, the feature element fitting method and the computer system thereof can quickly extract the uppermost point and the effective boundary point of the point cloud in the three-dimensional coordinate system, and use the iterative method to complete the fitting of the feature elements.

Referring to FIG. 1, a functional module diagram of a preferred embodiment of a computer system 1 for fitting feature elements of the present invention is shown. The computer system 1 can be divided into a receiving module 10, a triangular meshing module 12, an extracting module 14, a fitting module 16, and a creating module 18 according to the functions of the software block.

The receiving module 10 is configured to receive point cloud data generated by the image measuring machine, and the point cloud data includes the number of point clouds, the identifier of the point, and the attribute of each point.

The triangular meshing module 12 is configured to establish a triangular mesh surface according to the received point cloud data.

The extraction module 14 is configured to extract from the triangular mesh surface There are boundary points and the boundary points are extracted from the point cloud where the feature elements need to be fitted. Specifically, when the user selects a point cloud from the triangular mesh surface to fit the feature element by using the polyline, the extraction module 14 needs to find the uppermost point displayed on the screen from the selected point cloud, and then A boundary point is extracted from the topmost point for fitting a feature element. As shown in FIG. 2, it is a schematic diagram of the perspective normal vector (to be mentioned later) and the uppermost layer of the present invention. The figure is only used for illustration, and the actual arrangement of the point cloud is not limited thereto. In Fig. 2, reference numeral 2 denotes a screen of the display, reference numeral 3 denotes the uppermost layer, and reference numeral 4 denotes a view angle vector, as can be seen from Fig. 2, the uppermost layer is located at the uppermost layer of the viewing angle normal vector.

The fitting module 16 is configured to fit the boundary points in the uppermost layer into feature elements according to an iterative method, and obtain parameters of the fitted feature elements.

The creation module 18 is configured to draw feature elements according to the parameters of the feature element and the fitted feature elements.

Referring to Fig. 3, there is shown a flow chart of a preferred embodiment of the method for fitting a feature element of the present invention.

In step S1, the receiving module 10 receives the point cloud data generated by the image measuring machine to measure the workpiece, and the triangular meshing module 12 creates a triangular mesh surface according to the received point cloud data. The point cloud data includes the number of point clouds, the identifier of the points, and the attributes of each point.

In step S3, the extraction module 14 selects a point cloud from which the feature element needs to be fitted from the triangular mesh surface, and extracts a boundary point of the uppermost layer point from the selected point cloud.

In step S5, the fitting module 16 adopts an iterative method to the uppermost layer. The boundary points are fitted to feature elements. In the preferred embodiment, the iterative method is a quasi-Newton iteration method. Wherein, the feature element comprises a line, a face, a circle, a cylinder or a ball.

In step S7, the creating module 18 acquires parameters of the fitted feature elements, and draws the feature elements according to the parameters of the feature elements and the fitted feature elements. Wherein, when the feature element is a line, the parameter includes a starting point, an ending point, and a normal of the line; when the feature element is a face, the parameter includes a center point of the face and a normal of the face; When the feature element is a circle, the parameter includes a center of the circle, a radius of the circle, and a normal direction of the circle; when the feature element is a cylinder, the parameter includes a center point, a radius, a height, and a normal direction of the cylinder; the feature element When the ball is a ball, the parameters include the radius of the center of the ball and the ball.

Referring to FIG. 4, a detailed flowchart of step S3 in FIG. 3 is shown.

In step S30, the extraction module 14 extracts all boundary points in the triangular mesh surface. In this embodiment, the following is an example of determining whether a point is a boundary point: the extraction module 14 obtains a triangle around the current judgment point and a vertex of each triangle from the triangular mesh surface, and sequentially determines that the vertex is currently determined. Whether the number of times occupied by the triangle around the point is greater than 1, and if the result of the determination is no, it is determined that the current judgment point is a boundary point.

In step S32, the extraction module 14 selects a point cloud from the triangular mesh surface to be fitted with the feature element by using a polyline. The selected point cloud is a point cloud in a three-dimensional space, that is, a point cloud including multiple levels.

Step S34: Acquire the topmost point displayed on the screen from the selected point cloud, that is, from the perspective of the screen display, the layer point is located at the uppermost layer of the selected point cloud. This acquisition method will be specifically described in FIG.

In step S36, the attributes of the uppermost layer are compared with the attributes of the boundary points in step S30, respectively, to find the boundary points in the uppermost layer. Specifically, if the attribute of a certain point "a" in the uppermost layer is the same as the attribute of a certain boundary point in step S30, it is determined that the point "a" is the boundary point.

Referring to FIG. 5, it is a refinement flowchart of step S34 in FIG.

Step S340, calculating a bounding box of the point cloud according to the three-dimensional coordinates of the point in the point cloud data, grouping the bounding boxes, and filling the identifier of each point in the point cloud data into the corresponding group. The side length of each unit small cube after the grouping can be calculated from the maximum side length of the triangle constituting the triangular mesh surface and the total number of point clouds. The maximum side length is a user-defined side length, which is approximately equal to the average value of the distance between adjacent points in the point cloud.

In step S341, the group in which the current point is located is acquired, and at least one group is expanded to the periphery, for example, 3*3 groups are expanded outward to obtain triangles around all points in the group.

Step S343, creating a ray for the current point according to the viewing angle normal vector of the screen. The viewing angle normal vector of the screen refers to a normal vector formed by the user's perspective when viewing the computer screen, and the viewing angle normal vector is perpendicular to the user's horizontal line of sight.

In step S345, the number of intersections of the ray and the triangle is counted.

In step S347, it is determined whether the number of intersections is equal to 1.

If the number of intersections is not equal to 1, it indicates that the ray of the current point intersects with the plurality of triangles, so the flow is directly ended; if the number of intersections is equal to 1, then step S349 indicates that the current point is the uppermost point. , the current point is added to the array of the topmost point.

The extraction module 14 loops back the points in each group and repeats steps S341 through S349 until all of the uppermost points are found to be added to the array.

Referring to FIG. 6, a detailed flowchart of step S5 in FIG. 3 is shown.

Step S50, the user inputs a fitting type according to the approximate shape of the boundary points extracted in step S3 in FIG. 3, and the fitting type includes a fitting surface, a fitting line, a fitting circle, a fitting cylinder, and a fitting. ball.

In step S51, the fitting module 16 determines whether the input type of the fit is a fitted line or a fitted circle. If yes, go directly to step S52; if no, go to step S53.

In step S52, the fitting module 16 first fits the boundary point into a face, and then projects the boundary point onto the face, and then proceeds to step S53.

In step S53, a corresponding iterative equation is obtained according to the type of fitting input in step S50. Using this iterative equation, the minimum distance from each boundary point to the pre-fitted figure can be calculated.

Specifically, when the fitting type is a fitted line, the iterative equation of the fitted line is , where ( x _i , y _i , z _i ) is the coordinate of the current boundary point, ( x ₀ , y ₀ , z ₀ ) is the coordinate of the first endpoint of the fitted line, α is the current boundary point and the first The angle between the line of one point and the fitted line, where n is the number of boundary points. When the fitting type is a fitted circle, the iterative equation of the fitted circle is Where ( x _i , y _i ) is the coordinate of the current boundary point, ( x ₀ , y ₀ ) is the centroid of the fitted circle, R is the radius of the fitted circle, and n is the number of boundary points. When the fitting type is a fitting surface, the iterative equation of the fitting surface is Where ( x _i , y _i , z _i ) is the coordinate of the current boundary point, and n is the number of boundary points. When the fitting type is a fitting cylinder, the iteration equation of the fitting cylinder is , where ( x _i , y _i , z _i ) is the coordinate of the current boundary point, ( x ₀ , y ₀ , z ₀ ) is the first point coordinate of the central axis of the fitted cylinder, α is the current boundary point and the The angle between the line connecting the first point and the central axis, R is the radius of the cylinder, and n is the number of boundary points. When the fit type is a fit ball, the iteration equation of the fit sphere is , where ( x _i , y _i , z _i ) is the coordinate of the current boundary point, ( x ₀ , y ₀ , z ₀ ) is the coordinate of the sphere center of the fit sphere, R is the radius of the sphere, and n is the boundary point The number.

In step S54, the total number of iterations m is input, and the total number m of the iterations is specified by the user. The preferred embodiment is specifically described by taking an iteration once as an example, and the user can also perform multiple iterations to achieve a more precise purpose. For example, this embodiment can use i to represent the number of iterations, that is, the first iteration.

In step S55, the number n of point clouds in the iteration is obtained according to the number of iterations i and the total number of the boundary points. In the preferred embodiment, m=3 is assumed, and when i=1, that is, the first iteration, a ratio of 10:1 is taken from the boundary point. Equally spaced equally, that is, n = total number of boundary points * 1 / 10 at this time, wherein the specific steps of uniformly taking points from the boundary point at a ratio of 10:1 are: first along the X axis, Y axis, Z The axis direction divides the bounding box of the boundary point equally, so that the bounding box is evenly divided into 10 small bounding boxes, and then the centers of the 10 small bounding boxes are respectively determined, and finally the 10 pieces are respectively taken out from the 10 The nearest point of the center of the small bounding box; when i=2, that is, the second iteration, the point is evenly spaced from the boundary point by a ratio of 10:5, that is, n=the total number of point clouds of the object to be tested *5/10; When i=3 is the 3rd iteration, all boundary points are iterated.

In step S56, an initial iteration parameter is obtained by using a least squares method. In this embodiment, the initial iteration parameter of the fitted line is the coordinates of the two endpoints of the fitted line, the initial iteration parameter of the fitted circle is the center coordinate, and the initial iteration parameter of the fitted sphere is the spherical center coordinate, The initial iteration parameter of the cylindrical shape is the coordinate of the two end points of the central axis of the cylinder, and the initial iteration parameter of the fitting surface is the spatial position and the normal vector of the fitting surface in the coordinate system.

Step S57, calculating a coordinate of a point in the pre-fitted graph by using a quasi-Newton iteration method according to the initial iteration parameter. Specifically, the iteration value of each boundary point is calculated first, and then the virtual point cloud position A1 of the boundary point is obtained, and then, according to the total number of iterations m, whether the virtual point cloud position A1 is pre-planned is determined. The coordinates of the point in the graph. For example, if the number of iterations input in step S54 is greater than 1, the virtual point cloud position A1 of the boundary point may be iterated as the initial point cloud position at the next iteration to calculate the virtual point cloud position A2, and In this way, the virtual point cloud position A' of the last iteration is calculated as the coordinates of the point in the pre-fitted graph; The number of iterations input in step S54 is equal to 1, then A'=A1. Wherein, when the number of iterations is equal to 1, the iterative value is the minimum distance from the boundary point to the pre-fitted graph; when the number of iterations is greater than 1, the iterative value includes the boundary point and the virtual point cloud at the first iteration The minimum distance between the positions A1, the minimum distance between the virtual point cloud position A1 at the first iteration and the virtual point cloud position at the next iteration, and the virtual point cloud position A' of the last iteration to the pre-fit The minimum distance of the graph.

Step S58, fitting the feature element according to the coordinates of each point in the pre-fitted graph and the type of fitting input.

Referring to FIG. 7, a specific operation flowchart of the quasi-Newton iteration method in step S57 of FIG. 6 is shown.

In step S600, the initial iteration parameter is substituted into the iterative equation f(x), and the iterative function value f(x) is calculated.

In step S602, the fitting module 16 determines whether the calculated iterative function value f(x) is smaller than the fitting precision FunX. The fitting precision FunX is a user-defined fitting precision, which can be input in advance in the computer system 1, which specifically refers to a point cloud in the pre-fitted figure and a corresponding boundary point extracted in step S3 in FIG. The degree of error to be reached.

If f(x) is not smaller than FunX, the process proceeds to step S604, and the falling direction of f(x) is calculated by a mathematical rule such as a quasi-Newton iteration method. The downward direction refers to a direction in which the value of f(x) becomes smaller, even if the distance from the boundary point to the pre-fitted figure becomes smaller.

In step S606, it is determined whether the falling direction exists.

If the falling direction exists, proceed to step S608 to calculate a boundary point. The distance f(x)' after the fitting step D is moved in the descending direction to the pre-fitted figure. The fitting step length D refers to the distance of the center of the pre-fitted figure each time based on the corresponding boundary point extracted in step S3 in FIG. 3, and the fitting step size D is a user-defined parameter and can be used in a computer. System 1 is pre-entered.

Specifically, the position of the boundary point after moving D in the descending direction is first calculated, and then the distance f(x)' of the boundary point to the pre-fitted figure is calculated by using the position. Wherein, the f(x)' is completely the same as the calculation method of f(x) in step S600, and only the used parameters are different, and the calculation can be completed by referring to step S600.

In step S610, it is determined whether the calculated f(x)' is smaller than f(x). If f(x)' is smaller than f(x), the process returns to step S608; if f(x)' is not less than f(x), the process returns to step S604.

In step S606, if the falling direction does not exist, indicating that the iterative function value f(x) is the calculated iteration value, then in step S612, the iteration value is output, and then the flow is ended. In this embodiment, the coordinates of each point in the pre-fitted graph can be calculated using the calculated iteration values and the coordinates of the corresponding boundary points.

In step S602, if f(x) is smaller than FunX, the process proceeds directly to step S612.

The present invention has been described above in terms of preferred embodiments, and is not intended to limit the invention. Any person skilled in the art will be able to make changes and refinements without departing from the spirit and scope of the invention, and the scope of the invention is defined by the scope of the appended claims.

Computer system for fitting feature elements ‧‧1

Receiving module ‧‧10

Triangular grid module ‧‧12

Extraction module ‧‧14

Fitting module ‧‧16

Create a module ‧‧18

Screen ‧‧2

Top level ‧‧3

Perspective method vector ‧‧4

Enter the triangle grid after the point cloud ‧‧S1

Extract boundary points from point clouds that need to fit feature elements ‧‧‧S3

Fitting feature elements ‧‧‧S5

Get the parameters of the fitted feature element and draw the feature element ‧‧‧S7

1 is a functional block diagram of a preferred embodiment of a computer system for fitting feature elements of the present invention.

2 is a schematic diagram of a view normal vector and an uppermost layer of the present invention.

Fig. 3 is a flow chart showing the operation of a preferred embodiment of the feature element fitting method of the present invention.

FIG. 4 is a detailed flowchart of step S3 in FIG. 3.

FIG. 5 is a detailed flowchart of step S34 in FIG. 4.

FIG. 6 is a detailed flowchart of step S5 in FIG. 3.

Figure 7 is a flow chart showing the specific operation of the quasi-Newton iteration method in step S57 of Figure 6.

Enter the triangle grid after the point cloud ‧‧S1

Extract boundary points from point clouds that need to fit feature elements ‧‧‧S3

Fitting feature elements ‧‧‧S5

Get the parameters of the fitted feature element and draw the feature element ‧‧‧S7

Claims (10)

A feature element fitting method, comprising: receiving point cloud data from an image measuring machine; establishing a triangular mesh surface according to the received point cloud data; and selecting a point to be fitted with the feature element from the triangular mesh surface a cloud; obtaining an uppermost point of the screen from the selected point cloud; extracting a boundary point in the uppermost point; fitting the boundary point in the uppermost point to a feature element by using an iterative method; acquiring the fitted a parameter of the feature element; and drawing the feature element according to the parameter of the feature element and the fitted feature element. The feature element fitting method according to claim 1, wherein the step of obtaining an uppermost layer of the screen from the selected point cloud comprises the following steps: (a) calculating a point according to the point cloud data. a bounding box of the cloud, grouping the bounding boxes, and filling in the identifiers of each point in the point cloud data into the corresponding group; (b) acquiring the group in which the current point is located, and expanding at least one group to each surrounding area, To obtain a triangle around all points in the group; (c) to create a ray for the current point according to the screen view normal vector; (d) to count the number of intersections of the ray with the triangle; (e) when the number of intersections is equal to 1 Adding the current point to the array of the topmost point; and (f) Loop back the points in each group and repeat steps (b) through (e) until all the top points are found. The feature element fitting method according to claim 1, wherein after the step of creating a triangular mesh surface according to the point cloud data, the step of extracting the boundary point in the uppermost layer further comprises the following steps: extracting the All the boundary points in the triangle mesh surface. The feature element fitting method according to claim 3, wherein the step of extracting a boundary point in the uppermost layer point comprises the step of: locating an attribute of the uppermost layer point and the triangular mesh surface The attributes of all the boundary points are compared separately to find the boundary points in the uppermost point. The feature element fitting method according to claim 1, wherein the step of fitting the boundary point in the uppermost point to the feature element by using an iterative method comprises the following steps: (i) inputting a fitting type, The type of fit includes a fitting surface, a fitted line, a fitted circle, a fitted cylinder, and a fitted sphere; (ii) determining whether the type of fit entered is a fitted line or a fitted circle; (iii) if yes, Fitting the boundary point into a face and then projecting the boundary point onto the face; (iv) obtaining an initial iteration parameter by least squares according to the type of fit input; (v) initial iteration parameter Substituting into the iterative equation of the input fitting type, and using the quasi-Newton iteration method to calculate the coordinates of the points in the pre-fitted graph; (vi) fitting the feature elements according to the coordinates of each point in the pre-fitted graph and the type of fit entered. The feature element fitting method according to claim 5, wherein the step (v) comprises the following steps; (v1) substituting the initial iteration parameter into the iterative equation f(x) of the input fitting type, and calculating The iterative function value f(x); (v2) when the calculated iterative function value f(x) is not less than the fitting precision FunX, proceeds to step (v3); (v3) uses the quasi-Newton iteration method to calculate f(x) a direction of descent, which refers to a direction in which the distance from the boundary point to the pre-fitted figure becomes smaller; (v4) when there is the descent direction, the distance from the pre-fit figure after the boundary point is moved in the descending direction by the fitting step length is calculated f(x)', wherein the fitting step refers to the distance of the center of the pre-fitted figure each time based on the extracted boundary point; (v5) when f(x)' is smaller than f(x), Returning to step (v4), if f(x)' is not less than f(x), it returns to step (v3). The feature element fitting method according to claim 6, wherein when f(x) is smaller than FunX or when the descending direction does not exist, the method further comprises the following steps: an iterative function value f(x) For the calculated iteration value, the coordinates of each point in the pre-fitted graph are calculated from the iteration value and the coordinates of the boundary point. A computer system for fitting a feature element, the computer system comprising: a triangular meshing module for establishing a point cloud into a triangular mesh surface; and an extraction module for selecting a fitting from the triangular mesh surface Feature element a point cloud, obtaining an uppermost point of the screen from the selected point cloud, and extracting a boundary point in the uppermost point; a fitting module for using an iterative method to set a boundary in the uppermost point The points are fitted into the feature elements, and the parameters of the fitted feature elements are obtained; and a module is created for drawing the feature elements according to the parameters of the feature elements and the fitted feature elements. The computer system of claim 8, wherein the extraction module is further configured to extract all boundary points in the triangular mesh surface, and attributes of the uppermost layer point and attributes of the all boundary points A comparison is made separately to find the boundary points in the uppermost point. The computer system of claim 8, wherein the extracting module performs the following steps in the process of obtaining the topmost point of the screen from the selected point cloud: (a) calculating according to the point cloud data Obtaining a bounding box of a point cloud, grouping the bounding boxes, and filling in the identifiers of each point in the point cloud data into the corresponding group; (b) acquiring the group in which the current point is located, and expanding at least one of the surrounding points Grouping to obtain triangles around all points in the group; (c) creating a ray for the current point based on the screen view normal vector; (d) counting the number of intersections of the ray with the triangle; (e) when the number of intersections When equal to 1, the current point is added to the array of the topmost point; and (f) the points in each group are looped, and steps (b) through (e) are repeated until all uppermost points are found.