CN105574812B - Multi-angle three-dimensional data method for registering and device - Google Patents

Abstract

The present invention discloses a multi-angle three-dimensional data registration method and device, wherein the method includes: acquiring multiple sets of three-dimensional data for each rotation of a preset angle on a turntable of a three-dimensional scanning system by a reference plane; the reference plane is perpendicular to the turntable; Each set of three-dimensional data is fitted to obtain multiple fitting planes; according to the direction of the intersection line between the multiple fitting planes and the same section, the rotation direction of the rotating shaft of the three-dimensional scanning system is determined; according to the spatial coordinates of the pivot point on the rotating shaft and the The positional relationship between multiple fitting planes and intersection lines of the same section determines the optimal spatial coordinates of the rotating shaft; according to the optimal spatial coordinates of the rotating shaft, the 3D scanning data of the object to be scanned acquired under the 3D scanning system is registered. The above technical solution improves the efficiency and accuracy of multi-angle three-dimensional data registration.

Description

Multi-angle three-dimensional data registration method and device

technical field

The invention relates to the technical field of data registration, in particular to a multi-angle three-dimensional data registration method and device.

Background technique

The existing 3D scanning system can only scan the visible information on the surface within the scanning range. In order to obtain a complete 3D model of the object, it is often necessary to move the scanner itself, or rotate or move the scanned object to complete the object. Acquisition of angular three-dimensional information. Due to changes in the position of the scanning system or the position of the object, each scan data is limited to a different scanner coordinate system or object coordinate system, resulting in that the output 3D data cannot be directly and orderly arranged in a unified coordinate system. In order to obtain the complete 3D information of the object, it is necessary to match these multi-angle 3D scanning data. In existing 3D scanning systems, there are four commonly used scan data registration methods as follows:

(1) Manual method: Manually select the corresponding points on the adjacent scan data, and then calculate the transformation matrix according to the position transformation of the corresponding points to register the scan data. This type of method requires a large overlapping area of adjacent scanning data, which limits the scanning rotation angle, resulting in a decrease in scanning efficiency. Another disadvantage of this type of method is that the accuracy of data arrangement depends on the input accuracy of the operator, and for surfaces without obvious features, the input accuracy is even more difficult to guarantee. Again, this type of non-automatic registration method is labor-intensive and inefficient.

(2) Marking method: A small detectable object is attached to the scanned object to become a mark or target, which is recognized by the operator and input accurately or through other auxiliary image processing algorithms and geometric algorithms to identify corresponding points for input , to complete the registration of adjacent scan data. Compared with the manual method, this type of method has improved accuracy and efficiency, but still has the following disadvantages: ① It must be ensured that there are a certain number of marking points in the line of sight of two adjacent scans, thus limiting the number of scanned objects each time. The rotation angle increases the number of scans and the amount of data; ② Marking or adding marks on the object may cause damage to the scanned object. At the same time, the physical size of the mark may cause the surface information of the scanned object to be covered, resulting in details of the scanned data. loss; ③The number and size of markers cannot establish a unified standard. If the number of markers is too small, there may be insufficient markers in the scanned data. If the number of markers is too large, it may cause serious interference in the scanned data. detection, the marker physical size is too large, which may result in less accurate operator input.

(3) Image and geometric analysis methods: This type of method completes the registration of the model by detecting images or scan data of adjacent perspectives, detecting and extracting features, and registering the features at the same time. The disadvantages of this type of method are: ① the method has limitations on the surface features and texture of the scanned object, and the method may not be able to complete the registration task when the texture is not obvious or the feature is not obvious; ② the method needs to detect the correspondence between adjacent images Pixels or detection of corresponding feature points of adjacent scanning data, the algorithm has a large amount of calculation, and the source of error cannot be guaranteed; ③Through the method of pixel matching and geometric feature matching, a certain overlapping area of adjacent scanning data is required to detect corresponding features, As a result, the angle of movement of the scanning viewing angle is limited, the number of scans is increased, and the scanning efficiency is reduced.

(4) Turntable axis marking method: by calculating the coordinates of the turntable axis in the scanner coordinate system, and at the same time calculating the transformation coordinates of each piece of scanned data by obtaining the rotation angle, so as to complete the overall data registration. The problem with this type of method at present is that it is impossible to accurately calibrate the rotation axis of the turntable to the scanner coordinate system. Usually, this type of calibration uses a sphere for calibration. However, due to the limited viewing angle of the scanner, the area of the sphere is limited each time it is scanned. The center of the calibration sphere is calculated by fitting, and the algorithm itself has a fitting error. At the same time, the ball is moved through the turntable to obtain three ball center positions, and the fitting is performed again to calculate the position of the turntable axis. After obtaining the rough matching data through the calculated axis, the data is further finely registered. This type of method cannot be separated from the fine registration step of the scan data, and the registration efficiency is low, and the applicability of various existing fine registration algorithms is limited, and it is impossible to use one algorithm to complete the fine registration of all models, and because the algorithm is stable Sexual problems, human judgment is required after each registration, which is labor-intensive. Another method is to place a plane object in the center of the rotating shaft, and use the intersection line of the plane data obtained through multiple scans as the rotating shaft, but the position of the plane object needs to be placed very precisely, and there are high operating requirements, so it is difficult to guarantee success every time. .

With the wide application of 3D digital technology, the existing 3D scanning technology is developing in the direction of automation, which is to automatically scan objects in 3D from various angles, and then realize the automatic registration process of multi-angle 3D data. The key issue involved is how to accurately and easily realize the automatic registration of multi-angle 3D information. At present, the most widely used is to use the precision turntable to control the rotation of the object, and then realize the multi-angle 3D information registration of the object according to the known rotation angle. allow. Examples similar to the registration of multi-angle 3D information in the prior art mainly include the following items:

(1) Chinese patent application No. 200410031157 discloses a method for automatic registration based on large-scale scene multi-viewpoint laser scanning data. The 3D feature reflects the inherent structural features of the scene, simplifies the geometric complexity of the scene, and is an effective primitive for multi-view 3D data registration. The invention ensures that the Z-axis of the laser rangefinder is vertical to the ground, and the X and Y axes are parallel to the ground to obtain scanning data, and at the same time ensures that the scanning data of adjacent viewpoints maintains an overlap of 10%-20%; extracts the intrinsic value of the measured scene Structural features simplify the geometric complexity of the scene; coarse registration, fine registration and global registration are performed on the data by calculating virtual features and constructing feature units. The invention is suitable for automatic registration of multi-viewpoint laser scanning data of modern structures.

(2) Chinese Patent Application No. 03817891 discloses a device and method for automatically arranging 3D scanning data using optics, wherein non-contact type markers are used to automatically arrange 3D data without losing or damaging objects part being scanned. The device uses optical markers for automatically aligning 3D scan data obtained by photographing an object at different angles, and includes: a marker generating device for projecting a plurality of optical markers onto the surface of an object, a blob projection device for A pattern is projected on the surface of the object to obtain 2D image data comprising the object projected on the surface of the object, and for obtaining 3D scan data of the object through the pattern projected on the surface of the object; the image obtaining device uses for obtaining 2D image data of the object comprising projected on the surface of the object, and for obtaining 3D scan data of the object by means of a pattern projected on the surface of the object; and control means for obtaining from the 2D image data The relationship between the marker and the 3D scan data calculates the 3D position of the marker, and calculates the relative position of the 3D scan data based on the 3D position of the marker.

(3) Chinese patent application No. 200710018782 discloses a method for automatic registration of multi-view 3D face scan data, which is characterized in that it includes the following steps: transform the multi-view 3D face scan data with the PCA method for coordinate axis transformation, and convert The final model calculates its Shape Index value, and uses the method of threshold segmentation to screen out several feature areas; uses the relative distribution characteristics of the area to constrain the screening of the area, and finally locates the sign area; the inner corner of the glasses, the outer corner of the eye and the tip of the nose area; for the sign The area is registered using the ICP method, and the transformation parameters obtained by the registration of the overall scan data and the Ayong area are subjected to coordinate evaluation, movement and rotation transformation to complete the registration. Since the coarse registration of the present invention adopts the coordinate axis conversion method, the overall feature is used instead of the individual point feature, which reduces the error caused by the feature point monitoring; the fine registration only performs the registration iterative operation on the proposed feature area, which reduces the The computational complexity realizes the automatic registration of multi-attitude 3D scanning data.

(4) Chinese Patent Application No. 200710122787 discloses an automatic mechanism for identifying and defining a global coordinate system that is most suitable for the collection of raw 3D scan data forming a mesh model. More specifically, the identified coordinate system minimizes the sum of bias errors while minimizing the peak error of the original 3D scan data. The invention searches for a suitable coordinate system from raw 3D scan data in a fully automatic manner. Multiple coordinate systems that minimize the overall bias error are identified and presented to the user. It also allows the end user to interactively edit the proposed coordinate system parameters before transforming the alignment of the 3D scan data based on the selected coordinate system.

(5) Chinese patent application No. 200810224183 discloses a method for automatic registration of depth images combined with texture information, which is used for reconstruction of 3D models of various real objects. The steps are: ① Extract from scan data or generate textures based on depth images Image; ②Extract the pixel of interest in the texture image based on SIFT features, and find out the candidate set of matching pixel pairs through the method of cross-checking; ③Find out the correct matching pixel pair in the candidate set according to the geometric information constraints; ④In three-dimensional space Find the matching vertex pairs corresponding to the matching pixels, and calculate the body replacement matrix between the two depth images; ⑤Use the improved ICP algorithm to optimize this result; ⑥Based on the registration of two depth images, the input of multi-depth images The sequence is divided into several strip-shaped subsequences; ⑦Using a forward search strategy to merge these subsequences and construct a complete 3D model. The invention can be used to complete large-scale three-dimensional scanning data and generate three-dimensional models.

(6) Chinese patent application No. 02138210.7 discloses a method for realizing multi-view data acquisition and alignment reset of a three-dimensional camera. The method of aligning and relocating the data collected by each viewing angle in the same space coordinate system, the method is to use a single-view 3D camera and a rotating table, place the rotating table within the effective viewfinder range in front of the 3D camera, Place a small ball on the rotating table, and use a single-view 3D camera to calibrate the parameters of the small ball on the rotating table; collect multi-view data on the surface of the measured object, and finally align and reset the collected data, and output the complete surface of the measured object three-dimensional data. The present invention can use a single-view three-dimensional camera to collect multiple viewing angles, and automatically align and reset the three-dimensional data collected by each viewing angle.

From the above, in order to realize the automatic registration of data after multi-angle 3D scanning, the 3D scanning method based on the precision rotating platform places the object on the turntable, and controls the rotation angle through the computer to realize the acquisition of 3D information from different angles of the object. , and then combine the rotation angle of the turntable to realize the automatic splicing of multi-angle 3D data. The main problem involved in this method is the calculation of the rotation axis of the turntable, that is, to determine the spatial position of the rotation axis relative to the 3D scanning device. At present, the commonly used method is to calculate its precise axis center and axis by fitting methods such as spheres or planes. position, but the accuracy of this method depends on the accuracy of the object placement. If the placement position deviates from the axis of the turntable, it is difficult to accurately calculate the rotation axis, and often requires multiple operations. The result is determined by empirical judgment and manual observation. Satisfied, the operation is more cumbersome.

Therefore, in the existing multi-angle 3D scanning data automatic registration scheme, the method for determining the spatial parameters of the rotating shaft in the 3D scanning system is complicated to operate, and the efficiency and accuracy are low, which also leads to the multi-angle 3D scanning data automatic registration operation is complex and inefficient. and low precision.

Contents of the invention

An embodiment of the present invention provides a multi-angle three-dimensional data registration method to improve the efficiency and accuracy of multi-angle three-dimensional data registration. The method includes:

Obtain multiple sets of 3D data for each preset angle of rotation of the reference plane on the turntable of the 3D scanning system; the reference plane is perpendicular to the turntable;

Fit each set of three-dimensional data to obtain multiple fitting planes;

Determine the rotation direction of the rotating shaft of the three-dimensional scanning system according to the directions of the intersection lines between the multiple fitting planes and the same section;

Determine the optimal spatial coordinates of the rotating shaft according to the spatial coordinates of the pivot point on the rotating shaft and the positional relationship between the multiple fitting planes and the intersection lines of the same section;

According to the optimal spatial coordinates of the rotating shaft, the three-dimensional scanning data of the object to be scanned acquired under the three-dimensional scanning system are registered.

On the other hand, an embodiment of the present invention provides a multi-angle three-dimensional data registration device to improve the efficiency and accuracy of multi-angle three-dimensional data registration. The device includes:

The three-dimensional data acquisition unit is used to acquire multiple sets of three-dimensional data of each preset angle of rotation of the reference plane on the turntable of the three-dimensional scanning system; the reference plane is perpendicular to the turntable;

The three-dimensional data fitting unit is used for fitting each group of three-dimensional data to obtain multiple fitting planes;

The rotation axis direction determination unit is used to determine the rotation direction of the rotation axis of the three-dimensional scanning system according to the direction of the intersection line between multiple fitting planes and the same section;

The rotating shaft spatial coordinate determination unit is used to determine the optimal spatial coordinates of the rotating shaft according to the spatial coordinates of the pivot point on the rotating shaft and the positional relationship between multiple fitting planes and the intersection lines of the same section;

The data registration unit is configured to register the three-dimensional scanning data of the object to be scanned acquired under the three-dimensional scanning system according to the optimal spatial coordinates of the rotating shaft.

Compared with the prior art, the technical solution provided by the embodiment of the present invention, when determining the spatial parameters of the three-dimensional scanning rotating shaft, does not need to place the reference objects for calculating the rotating shaft accurately, but roughly places them, which reduces the complexity of the operation; using a Simply perpendicular to the reference plane of the turntable, it is not required to be strictly placed in the center of the turntable, and multiple sets of 3D data are obtained for each preset angle of rotation of the reference plane on the turntable, and each set of 3D data is simulated by means of plane fitting. According to the direction of the intersection line between the multiple fitting planes and the same section, determine the rotation direction of the rotating shaft of the three-dimensional scanning system; according to the spatial coordinates of the pivot point on the rotating shaft and the multiple fitting planes and The positional relationship of the intersection lines of the same section can accurately solve the optimal spatial coordinates of a center point (axis point) on the rotating shaft, and then completely determine the optimal spatial coordinates of the rotating shaft, which improves the ability to determine the spatial parameters of the rotating shaft of the 3D scanning system Efficiency and accuracy, to ensure subsequent accurate multi-angle 3D data registration; finally, according to the optimal spatial coordinates of the rotating shaft, the 3D scan data of the object to be scanned acquired under the 3D scanning system is registered, therefore, The efficiency and accuracy of multi-angle 3D data registration are improved.

Description of drawings

The drawings described here are used to provide further understanding of the present invention, constitute a part of the application, and do not limit the present invention. In the attached picture:

FIG. 1 is a schematic flowchart of a multi-angle three-dimensional data registration method in an embodiment of the present invention;

2 is a schematic structural diagram of a three-dimensional scanning system in an embodiment of the present invention;

Fig. 3 is a schematic diagram of establishing a linear optimization function and solving the optimal rotation axis space parameters in the embodiment of the present invention;

Fig. 4 is a schematic structural diagram of a multi-angle three-dimensional data registration device in an embodiment of the present invention.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be described in further detail below in conjunction with the embodiments and accompanying drawings. Here, the exemplary embodiments and descriptions of the present invention are used to explain the present invention, but not to limit the present invention.

In order to realize the automatic registration of data after multi-angle 3D scanning, a 3D scanning method based on a precision rotating platform has emerged. The object is placed on the turntable, and the rotation angle is controlled by a computer to achieve 3D information acquisition from different angles of the object. Combining the rotation angle of the turntable to realize the automatic splicing of multi-angle 3D data. The main problem involved in this method is the calculation of the rotation axis of the turntable, that is, to determine the spatial position of the rotation axis relative to the 3D scanning device. At present, the commonly used method is to calculate its precise axis center and axis by fitting methods such as spheres or planes. s position.

However, the inventor found that the accuracy of this method depends on the accuracy of the placement of the object. If the placement deviates from the axis of the turntable, it is difficult to accurately calculate the rotation axis, and often requires multiple operations. It is determined by empirical judgment and manual observation. Whether the result is satisfactory or not, the operation is more cumbersome.

In order to solve this problem, the present invention proposes a multi-angle three-dimensional data registration method. When determining the spatial parameters of the three-dimensional scanning rotating shaft, the method does not need to accurately place the rotating shaft, and calculates the optimal calculation method of the reference object. First, a Planar objects are placed on the turntable, and by controlling the turntable to rotate 4-5 times, each rotation is 10-20 degrees, 4-5 sets of 3D data of the plane are obtained, and the space equations of these sets of planes are obtained through plane fitting, and each The direction of the intersection line between the planes, because the reference plane is perpendicular to the turntable, so the direction of this group of intersection lines is the direction of the rotation axis. In order to determine the position of the rotation axis, we also need to determine the position of a spatial point on the rotation axis. Through simple geometric analysis , we can easily find that the distance from each point on the rotation axis to this group of plane intersection lines is equal. Based on this phenomenon, we can establish a simple optimization function to find the center position of the rotation axis with the smallest error, and then determine the rotation axis at The precise position under the reference system of the 3D scanning system, and then each set of scanned data can be accurately registered according to its predetermined rotation angle. The proposed method is simple in operation, low in computational complexity, and accurate in results. It can avoid the complex iterative process (ICP, [1-2]) adopted by existing methods such as the ICP algorithm, and can accurately obtain the position of the rotating shaft and achieve fine global accuracy. Automatic registration operation.

The working diagram of the entire 3D scanning system is that the object is placed on a precision rotating platform, its rotation is controlled by a computer, and 3D information of different angles is obtained through a 3D scanner, and then the subsequent registration of multi-angle 3D information is completed.

Assuming that the actual rotation axis L of the turntable is defined by a point A(a,b,c) and a vector V(u,v,w), the rotation matrix for rotating θ around the axis is:

In order to realize the automatic registration of multiple sets of 3D scanning data, we only need to solve the point A and the direction vector V.

Therefore, first of all, we place an L-shaped plane on the turntable (it can also be other types of planes, as long as the plane is perpendicular to the plane of the turntable), so as to ensure that the plane is perpendicular to the plane of the turntable, and drive the turntable to rotate multiple times to obtain different Turn to the plane 3D data under the angle, perform fitting, and then obtain multiple sets of plane intersection lines, calculate the average of multiple sets of intersection line direction parameters, and then obtain the rotation axis direction parameter V;

Secondly, assuming that point A(a,b,c) is the coordinates of a point on the rotation axis, the intersection lines from each group of fitting planes should have the characteristic of equal distance, so a simple linear optimization function can be established to solve A The best position of the shaft, and then the spatial parameter information of the rotating shaft can be completely determined.

Firstly, the multi-angle 3D data registration method will be introduced in detail below.

Fig. 1 is a schematic flowchart of a multi-angle three-dimensional data registration method in an embodiment of the present invention, and Fig. 2 is a schematic structural diagram of a three-dimensional scanning system in an embodiment of the present invention, as shown in Figs. 1 and 2, the method includes the following steps:

Step 101: Obtain multiple sets of three-dimensional data for each rotation of the reference plane at a preset angle on the turntable of the three-dimensional scanning system; the reference plane is perpendicular to the turntable;

Step 102: Fitting each set of three-dimensional data to obtain multiple fitting planes;

Step 103: Determine the rotation direction of the rotating shaft of the three-dimensional scanning system according to the directions of intersection lines between multiple fitting planes and the same section;

Step 104: Determine the optimal spatial coordinates of the rotating shaft according to the spatial coordinates of the pivot point on the rotating shaft and the intersection lines between multiple fitting planes and the same section;

Step 105: According to the optimal spatial coordinates of the rotating shaft, register the 3D scanning data of the object to be scanned acquired under the 3D scanning system.

Compared with the prior art, the technical solution provided by the embodiment of the present invention, when determining the spatial parameters of the three-dimensional scanning rotating shaft, does not need to place the reference objects for calculating the rotating shaft accurately, but roughly places them, which reduces the complexity of the operation; using a Simply perpendicular to the reference plane of the turntable, it is not required to be strictly placed in the center of the turntable, and multiple sets of 3D data are obtained for each preset angle of rotation of the reference plane on the turntable, and each set of 3D data is simulated by means of plane fitting. According to the direction of the intersection line between the multiple fitting planes and the same section, determine the rotation direction of the rotating shaft of the three-dimensional scanning system; according to the spatial coordinates of the pivot point on the rotating shaft and the multiple fitting planes and The positional relationship of the intersection lines of the same section can accurately solve the optimal spatial coordinates of a center point (axis point) on the rotating shaft, and then completely determine the optimal spatial coordinates of the rotating shaft, which improves the ability to determine the spatial parameters of the rotating shaft of the 3D scanning system Efficiency and accuracy, to ensure subsequent accurate multi-angle 3D data registration; finally, according to the optimal spatial coordinates of the rotating shaft, the 3D scan data of the object to be scanned acquired under the 3D scanning system is registered, therefore, reducing The operation complexity of multi-angle three-dimensional data registration is reduced, and the efficiency and accuracy of multi-angle three-dimensional data registration are improved.

The three-dimensional data mentioned in the embodiments of the present invention may be three-dimensional point cloud data.

The following describes in detail the acquisition and processing of the reference plane data in the above step 101:

Place the reference plane on the turntable without requiring it to pass through the center of the turntable;

The specific operation can be: control the turntable to rotate 4-5 times, each rotation is 10-20 degrees, and obtain 3D data of 4-5 sets of planes.

The following describes the fitting of each set of three-dimensional data in the above step 102 to obtain multiple fitting planes:

After obtaining the 3D data at different angles of the reference plane, before fitting each set of 3D data, other operations that can be performed include: denoising the obtained point cloud data of multiple fitting planes, using The method of least squares fitting is used to obtain the equations of each plane as follows:

A _i x+B _i y+C _i z+D _i ＝0 (1)

The calculation of the rotating shaft direction in the above-mentioned step 103 is introduced below:

Since the reference plane is perpendicular to the table surface of the turntable, we can obtain the rotation axis direction by calculating the intersection of several sets of fitting planes:

Since the direction of the rotation axis has been determined, we use the obtained direction of the rotation axis as a normal vector to construct a new plane:

n _x x + n _y y + n _z z = 0 (3)

Then the newly constructed plane cuts each point cloud plane as shown in Figure 3, and the equation of the intersection line l _i of the new plane cut point cloud plane shown in Figure 3 is expressed as:

In one embodiment, the above step 104 may include:

Find multiple pivot points, and calculate the error between the distances from each pivot point to the intersection line of multiple fitting planes and the same section; The error between the distances of the lines, to find the optimal pivot point space coordinates;

According to the optimal spatial coordinates of the pivot point, the optimal spatial coordinates of the rotating shaft are determined.

In one embodiment, multiple pivot points are found, and the errors between the distances from each pivot point to multiple fitting planes and intersection lines of the same section are calculated respectively; according to each pivot point to multiple fitting The error between the distance between the plane and the intersection line of the same section, to find the optimal pivot point space coordinates, can include:

According to the linear optimization function, iteratively find the optimal pivot point spatial coordinates, and perform the following operations for each iteration cycle:

Calculate the distance from each pivot point to the intersection of multiple fitted planes with the same section;

Computes the error between the distances from each pivot point to the intersections of multiple fitted planes with the same section until the error between the distances from the pivot point to the intersections of multiple fitted planes with the same section is found to be minimal , the spatial coordinates of the corresponding pivot point, as the optimal spatial coordinates of the pivot point;

The linear optimization function is established according to the positional relationship between the spatial coordinates of the pivot point on the rotating shaft and the intersection lines between multiple fitting planes and the same section.

In one embodiment, the linear optimization function can be:

Among them, d _i is the distance from the i-th intersection line to the pivot point, d _j is the distance from the j-th intersection line to the pivot point, O is the space coordinate of the pivot point to be obtained, and O _init is the initial Pivot space coordinates.

The calculation of the center point of the rotating shaft (pivot point) and the derivation process of the linear optimization function are described in detail below.

It has been analyzed above: the distance from the rotating shaft center (rotating shaft pivot point) O to each straight line is equal. Wherein, as shown in Figure 3, the distance from the center of the rotating shaft to the straight line l _i can be expressed as:

in:

Since the center of the rotation axis is on the plane of construction, then we can get:

Since the distance from the center of the rotating shaft to each straight line is equal, there are:

By transforming (7) into the form of matrix representation, we can get

Wherein the matrix S is the coefficient factor of the vector [O _x , O _y , O _z ] ^T. Using SVD (singular valuedecomposition method, singular value decomposition method) can obtain an initial rotation axis center. By pair:

Carrying out a simple linear optimization solution can obtain the precise coordinates of the center point of the rotating shaft. Knowing the direction of the rotating shaft and a certain point on it, the spatial parameters of the rotating shaft under the 3D scanning system can be accurately determined, and then according to the rotation angle of the turntable, the The 3D data of the subsequent scans are automatically registered precisely and simply.

Only two intersection lines l _i and l _j are schematically marked in Fig. 3. When actually calculating the error between the pivot point and the intersection line of each group of fitting planes, it is necessary to calculate all the intersection lines on the figure The error value for the distance from the pivot point.

During specific implementation, in the embodiment of the present invention, the spatial coordinates of the rotating shaft are determined according to the positional relationship between the spatial coordinates of the pivot point on the rotating shaft and the intersection lines between multiple fitting planes and the same section, and the above-mentioned linear optimization function can be used to determine the spatial coordinates of the rotating shaft to represent the above relationship. Of course, besides establishing the linear optimization function to iteratively find the optimal spatial coordinates of the rotating shaft, there are many ways, for example, by establishing a spatial coordinate based on the pivot point on the rotating shaft and multiple fitting The table of the positional relationship between the plane and the intersection line of the same section can be used to find the optimal space coordinates of the rotation axis.

In one embodiment, the turntable may be a numerically controlled turntable. Of course, the CNC turntable can also be replaced by a hand-controlled turntable, and it is only necessary to ensure the accuracy of the rotation angle during operation.

In an example, it may also include: performing denoising processing on multiple sets of acquired three-dimensional data;

Fitting each set of 3D data includes: fitting each set of 3D data after denoising processing. Such a design ensures the accuracy of determining the spatial parameters of the rotating shaft and the registration of the three-dimensional data.

Based on the same inventive concept, an embodiment of the present invention also provides a multi-angle three-dimensional data registration device, such as the following embodiment. Since the problem-solving principle of the multi-angle 3D data registration device is similar to that of the multi-angle 3D data registration method, the implementation of the multi-angle 3D data registration device can refer to the implementation of the multi-angle 3D data registration method, and the repetition will not be repeated. . As used below, the term "unit" or "module" may be a combination of software and/or hardware that realizes a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementations in hardware, or a combination of software and hardware are also possible and contemplated.

Fig. 4 is a schematic structural diagram of a multi-angle three-dimensional data registration device in an embodiment of the present invention. As shown in Fig. 4, the device includes:

A three-dimensional data acquisition unit 10, configured to acquire multiple sets of three-dimensional data for each preset angle of rotation of a reference plane on the turntable of the three-dimensional scanning system; the reference plane is perpendicular to the turntable;

A three-dimensional data fitting unit 20, configured to fit each set of three-dimensional data to obtain multiple fitting planes;

The rotation axis direction determination unit 30 is used to determine the rotation direction of the rotation axis of the three-dimensional scanning system according to the direction of the intersection of multiple fitting planes and the same section;

The rotating shaft spatial coordinate determination unit 40 is used to determine the optimal spatial coordinates of the rotating shaft according to the spatial coordinates of the pivot point on the rotating shaft and the positional relationship between multiple fitting planes and the intersection lines of the same section;

The data registration unit 50 is configured to register the three-dimensional scanning data of the object to be scanned acquired under the three-dimensional scanning system according to the optimal spatial coordinates of the rotation axis.

In one embodiment, the rotation axis space coordinate determining unit 40 can be specifically used for:

Find multiple pivot points, and calculate the error between the distances from each pivot point to the intersection line of multiple fitting planes and the same section; The error between the distances of the lines, to find the optimal pivot point space coordinates;

According to the optimal spatial coordinates of the pivot point, the optimal spatial coordinates of the rotating shaft are determined.

In one embodiment, multiple pivot points are found, and the errors between the distances from each pivot point to multiple fitting planes and intersection lines of the same section are calculated respectively; according to each pivot point to multiple fitting The error between the distance between the plane and the intersection line of the same section, find the optimal pivot point space coordinates, including:

According to the linear optimization function, iteratively find the optimal pivot point spatial coordinates, and perform the following operations for each iteration cycle:

Calculate the distance from each pivot point to the intersection of multiple fitted planes with the same section;

Computes the error between the distances from each pivot point to the intersections of multiple fitted planes with the same section until the error between the distances from the pivot point to the intersections of multiple fitted planes with the same section is found to be minimal , the spatial coordinates of the corresponding pivot point, as the optimal spatial coordinates of the pivot point;

The linear optimization function is established according to the positional relationship between the spatial coordinates of the pivot point on the rotating shaft and the intersection lines between multiple fitting planes and the same section.

In one embodiment, the linear optimization function can be:

Among them, d _i is the distance from the i-th intersection line to the pivot point, d _j is the distance from the j-th intersection line to the pivot point, O is the space coordinate of the pivot point to be obtained, and O _init is the initial Pivot space coordinates.

In summary, the beneficial technical effects of the technical solutions provided by the embodiments of the present invention are as follows:

(1) When determining the space parameters of the three-dimensional scanning rotating shaft, it is not necessary to place the reference object for calculating the rotating shaft accurately, but roughly place it, which reduces the complexity of the operation;

(2) It is not necessary to know the marked points on the reference plane, but to calculate based on the spatial equation of the entire reference plane. The error of the 3D reconstruction system itself will not have a significant impact on the fitting results of the entire reference plane, thus ensuring operational stability;

(3) By establishing a linear optimization function, the spatial coordinates of the center point (pivot point) on the rotating shaft are accurately solved, which ensures the accuracy of the results.

Obviously, those skilled in the art should understand that each module, device or each step of the above-mentioned embodiments of the present invention can be implemented by a general-purpose computing device, and they can be concentrated on a single computing device, or distributed among multiple computing devices alternatively, they may be implemented in program code executable by a computing device, whereby they may be stored in a storage device to be executed by a computing device, and in some cases may be executed in a format different from The steps shown or described here are performed sequentially, or they are fabricated into individual integrated circuit modules, or multiple modules or steps among them are fabricated into a single integrated circuit module. Thus, embodiments of the invention are not limited to any specific combination of hardware and software.

The above are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, various modifications and changes may be made to the embodiments of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (8)

1. A multi-angle three-dimensional data registration method, characterized in that, Obtain multiple sets of three-dimensional data for each rotation of the reference plane at a preset angle on the turntable of the three-dimensional scanning system; the reference plane is perpendicular to the turntable; Fit each set of three-dimensional data to obtain multiple fitting planes; Determine the rotation direction of the rotating shaft of the three-dimensional scanning system according to the directions of the intersection lines between the multiple fitting planes and the same section; Determine the optimal spatial coordinates of the rotating shaft according to the spatial coordinates of the pivot point on the rotating shaft and the positional relationship between the multiple fitting planes and the intersection lines of the same section; registering the three-dimensional scanning data of the object to be scanned acquired under the three-dimensional scanning system according to the optimal spatial coordinates of the rotating shaft; According to the positional relationship between the spatial coordinates of the axis point on the rotating shaft and the intersection lines between multiple fitting planes and the same section, determine the optimal spatial coordinates of the rotating shaft, including: Find multiple pivot points, and calculate the error between the distances from each pivot point to the intersection line of multiple fitting planes and the same section; The error between the distances of the lines, to find the optimal pivot point space coordinates; According to the optimal spatial coordinates of the pivot point, the optimal spatial coordinates of the rotating shaft are determined. 2. The method for multi-angle three-dimensional data registration as claimed in claim 1, wherein a plurality of pivot points are found, and the distance between each pivot point and the intersection line of multiple fitting planes and the same section is calculated respectively. According to the error between the distances between each pivot point and the intersection line of multiple fitting planes and the same section, find the optimal pivot point spatial coordinates, including: According to the linear optimization function, iteratively find the optimal pivot point spatial coordinates, and perform the following operations for each iteration cycle: Calculate the distance from each pivot point to the intersection of multiple fitted planes with the same section; Computes the error between the distances from each pivot point to the intersections of multiple fitted planes with the same section until the error between the distances from the pivot point to the intersections of multiple fitted planes with the same section is found to be minimal , the spatial coordinates of the corresponding pivot point, as the optimal spatial coordinates of the pivot point; The linear optimization function is established according to the positional relationship between the spatial coordinates of the pivot point on the rotating shaft and the intersection lines between multiple fitting planes and the same section. 3. multi-angle three-dimensional data registration method as claimed in claim 2, is characterized in that, described linear optimization function is: stO = O _init ; Among them, d _i is the distance from the i-th intersection line to the pivot point, d _j is the distance from the j-th intersection line to the pivot point, O is the space coordinate of the pivot point to be obtained, and O _init is the initial Pivot space coordinates. 4. The multi-angle three-dimensional data registration method according to claim 1, wherein the turntable is a numerical control turntable. 5. The multi-angle three-dimensional data registration method according to claim 1, further comprising: performing denoising processing on multiple sets of three-dimensional data acquired; Fitting each set of 3D data includes: fitting each set of 3D data after denoising processing. 6. A multi-angle three-dimensional data registration device, characterized in that it comprises: A three-dimensional data acquisition unit, configured to acquire multiple sets of three-dimensional data for each preset angle of rotation of the reference plane on the turntable of the three-dimensional scanning system; the reference plane is perpendicular to the turntable; The three-dimensional data fitting unit is used for fitting each group of three-dimensional data to obtain multiple fitting planes; The rotation axis direction determination unit is used to determine the rotation direction of the rotation axis of the three-dimensional scanning system according to the direction of the intersection line between multiple fitting planes and the same section; The rotating shaft spatial coordinate determination unit is used to determine the optimal spatial coordinates of the rotating shaft according to the spatial coordinates of the pivot point on the rotating shaft and the positional relationship between multiple fitting planes and the intersection lines of the same section; A data registration unit, configured to register the three-dimensional scanning data of the object to be scanned acquired under the three-dimensional scanning system according to the optimal spatial coordinates of the rotating shaft; The rotating shaft space coordinate determining unit is specifically used for: Find multiple pivot points, and calculate the error between each pivot point and the intersection line of multiple fitting planes and the same section; The error between the distances of the lines, to find the optimal pivot point space coordinates; According to the optimal spatial coordinates of the pivot point, the optimal spatial coordinates of the rotating shaft are determined. 7. The multi-angle three-dimensional data registration device as claimed in claim 6, characterized in that, after finding a plurality of pivot points, the distance between each pivot point and the intersection line of multiple fitting planes and the same section is calculated respectively. According to the error between each pivot point and the distance between multiple fitting planes and the intersection line of the same section, find the optimal pivot point spatial coordinates, including: According to the linear optimization function, iteratively find the optimal pivot point spatial coordinates, and perform the following operations for each iteration cycle: Calculate the distance from each pivot point to the intersection of multiple fitted planes with the same section; Computes the error between the distances from each pivot point to the intersections of multiple fitted planes with the same section until the error between the distances from the pivot point to the intersections of multiple fitted planes with the same section is found to be minimal , the spatial coordinates of the corresponding pivot point, as the optimal spatial coordinates of the pivot point; The linear optimization function is established according to the positional relationship between the spatial coordinates of the pivot point on the rotating shaft and the intersection lines between multiple fitting planes and the same section. 8. The multi-angle three-dimensional data registration device according to claim 7, wherein the linear optimization function is: stO = O _init ; Among them, d _i is the distance from the i-th intersection line to the pivot point, d _j is the distance from the j-th intersection line to the pivot point, O is the space coordinate of the pivot point to be obtained, and O _init is the initial Pivot space coordinates.