CN108596847A - Deep hole inner surface image geometric distortion correction method based on multi-line structured light - Google Patents

Abstract

The invention discloses a method for correcting the geometric distortion of deep hole inner surface images based on multi-line structured light, comprising the following steps: step 1, establishing a deep hole inner surface model and a corresponding deep hole inner surface unfolding model; Obtain the surface acquisition image for the deep hole inner surface model and the plane acquisition image for the deep hole inner surface expansion model; Step 3, design the Gaussian structured light lattice pattern; Step 4, continue to interpolate the Gaussian points in the same row or column; Step 5. Relying on the image position and deformation relationship between the corresponding Gaussian structured light points in the curved surface acquisition image obtained by scanning the inner surface of the deep hole with the Gaussian structured light lattice pattern and the plane acquired image, the image distortion is corrected; the multi-line based The geometric distortion correction method of deep hole inner surface image based on structured light can be used to correct the geometric distortion of the collected image of the inner surface of common deep hole parts. It has simple calculation, high precision, non-contact scanning and wide application range.

Description

A geometric distortion correction method for deep hole inner surface images based on multi-line structured light

technical field

The invention relates to a method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light, and belongs to the technical fields of digital image processing and machine vision.

Background technique

As an important part of modern industrial manufacturing, deep hole parts are widely used in the production process of oil and gas pipelines, artillery barrels and other components. For this reason, it is very necessary to regularly inspect the deep hole parts during use, and the most important inspection content is the three-dimensional inspection of the geometric shape of the inner surface of the deep hole. At present, the detection of the inner surface of deep hole parts is mainly to modify and apply the related methods for the detection of the outer surface of the measured object. Among them, the most widely used optical detection method is based on structured light. According to different light sources, this method can be divided into structured light projection method and laser projection method; considering the limited physical space on the inner surface of deep hole parts, structured light laser detection equipment is easier to integrate and compress, and is more suitable for deep hole inspection. 3D detection.

The detection method based on structured light laser is to project linear structured light laser stripes onto the surface of the measured object, use the camera to collect the structured light stripes deformed by the geometric shape modulation of the measured object surface, and analyze the degree of deformation of the stripes and the geometric shape of the measured object surface Inversely obtain the physical information of the change in the geometric shape of the surface of the measured object, and finally complete the size measurement and three-dimensional shape restoration for industrial deep hole parts.

But at present, in the process of detecting the geometric shape of the inner surface of the deep hole with structured light, it is mainly based on calculating the perpendicularity between the broken stripes after modulation without considering the error factors such as the radian of the inner surface of the deep hole and the variation of the groove depth. Image distance, and then deduce the depth information of the actual geometry. However, in the process of actually detecting the inner surface of deep holes, the modulated structured light images collected by the CCD camera have different degrees of geometric misalignment just because of the above error factors. The main manifestations of geometric dislocation are as follows: First, due to the radian characteristics of the inner surface of the deep hole, the structured light stripes in the image are modulated and present a curved shape, which directly leads to the linear fitting of the stripes, resulting in deviations in the results; second, the limitation Due to the limited space inside the deep hole, it is difficult for the structured light projection equipment to project the laser vertically onto the inner surface of the deep hole; the third is the distribution characteristics of the geometric shapes on the inner surface of the deep hole. Structured light needs to be projected obliquely at a certain angle, but this also causes the stripes in the image to appear diffuse; Fourth, with the continuous expansion of the camera's field of view, inner surface curvature, and groove depth, the greater the distortion of the collected image, for The lower the measurement accuracy of the geometric parameters of the inner surface is. In order to further improve the accuracy of structured light detection, correcting the geometric distortion of structured light images on the inner surface of deep hole parts has attracted more and more people's attention and participation.

Contents of the invention

In order to solve the above problems, the present invention proposes a method for correcting the geometric distortion of deep hole inner surface images based on multi-line structured light, which can be applied to the geometric distortion existing in the collected images of the inner surface of various types of deep hole parts. The calibration is simple in calculation, high in precision, non-contact scanning and widely applicable.

The geometric distortion correction method of deep hole inner surface image based on multi-line structured light of the present invention comprises the following steps:

Step 1, establishing a deep hole inner surface model and a corresponding deep hole inner surface expansion model;

Step 2, obtaining a curved surface acquisition image for the deep hole inner surface model and a plane acquisition image for the deep hole inner surface unfolded model;

Step 3, designing a Gaussian structured light lattice pattern;

Step 4, continue to interpolate the Gaussian points in the same row or column;

Step 5: Based on the image position and deformation relationship between the corresponding Gaussian structured light points in the curved surface acquisition image obtained by scanning the inner surface of the deep hole with the Gaussian structured light lattice pattern and the plane acquired image, the image distortion is corrected.

Further, the specific operation method of the first step is as follows: for common deep hole parts, use 3D MAX software to establish the deep hole inner surface model (DIM model); and then establish the corresponding deep hole inner surface expansion model.

Still further, the developed model of the inner surface of the deep hole in the first step is to obtain the developed plane model (DIPM model) of the inner surface of the deep hole by developing the inner surface of the inner surface model of the deep hole relative to the axis into a plane.

Further, the specific operation method of the second step is as follows: use the pre-set multi-line structured light stripe pattern to project obliquely onto the surface of the measured object, and at the same time use the camera to obliquely shoot the structured light scanning area to obtain the required acquisition image , specifically, in a software simulation environment, the structured light laser stripes are respectively projected onto the surfaces at the same position of the two types of corresponding models, and corresponding acquisition images are obtained respectively.

Further, the specific operation method of the third step is as follows: Arrange the structured light spots according to the Gaussian distribution in the form of a row-column matrix, and transform the arranged light spot density according to the complexity of the surface geometry of the measured object. The surface geometry is complex , the light spot density increases, otherwise the density decreases.

Further, the specific operation method of step 4 is as follows: according to the cubic spline interpolation algorithm, interpolate from the horizontal direction and fit all Gaussian points on a single line in each direction, and obtain the corresponding interpolated fitting curve in the two types of model acquisition images. Correspondence between.

Further, the specific operation method of Step 5 is as follows: project the Gaussian structured light lattice pattern onto the inner surface model of the deep hole and the planar model after the inner surface of the deep hole is unfolded, and obtain the corresponding curved surface acquisition image and plane acquisition image respectively ; Analyze the image position and the corresponding relationship between the Gaussian structured light points on the respective images, and then construct a complete deformation function curve through an interpolation algorithm, so as to know the correction of the geometric distortion contained in the surface acquisition image.

Further, in the step five, the curved surface acquisition image and the plane acquisition image are processed, and the processing method is as follows: an edge detection algorithm based on gray scale difference is used to preprocess the image, as shown in the following formula:

; (1)

Identify all Gaussian points in the image, and obtain the abscissa matrix Vx and ordinate matrix Vy that record the image position of the Gaussian points, as shown in the following formulas respectively:

; (2)

; (3)

The cubic spline difference algorithm is used to interpolate all the Gaussian structured light points in the same row; the image positions between the corresponding Gaussian structured light points in the two types of images are compared, and the mutual deformation relationship is calculated.

Furthermore, the image geometric correction method in the step five is as follows: relying on the image position relationship between the corresponding Gaussian points in the two types of images obtained, correct the surface acquisition image of the multi-line structured light acquired during the actual simulation process , to correct the original surface acquisition image; the transverse coordinate correction and longitudinal coordinate correction are shown in the following formulas respectively:

;(4)

. (5)

Compared with the prior art, the present invention is based on the method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light, and obtains its three-dimensional shape by scanning the deep hole inner surface with multi-line structured light fringe pattern; through the Gaussian structure The light lattice pattern scans the inner surface of the deep hole and the corresponding expansion model to determine the deformation law of the inner surface model; in the analysis of the expandable surface model, by using the cubic spline interpolation algorithm, interpolation from the horizontal direction, Fit all Gaussian points on a single line in each direction, and then construct the distortion model existing on the inner surface of the deep hole, which is convenient for subsequent comparison and correction.

Description of drawings

Fig. 1 is a schematic structural view of the deep hole inner surface model of the present invention.

Fig. 2 is a structural schematic diagram of the deep hole inner surface unfolding model of the present invention.

Fig. 3 is a schematic diagram of the structure of the curved surface acquisition image of the present invention.

Fig. 4 is a schematic diagram of the planar model structure of the present invention.

Fig. 5 is a schematic diagram of the Gaussian structured light structure of the present invention.

Fig. 6 is a schematic diagram of the Gaussian point surface acquisition image structure of the present invention.

Fig. 7 is a schematic diagram of the Gauss point plane acquisition image structure of the present invention.

Fig. 8 is a schematic diagram of the Gaussian point interpolation structure of the curved surface acquisition image of the present invention.

Fig. 9 is a schematic diagram of the Gaussian point interpolation structure of the plane acquisition image of the present invention.

Fig. 10 is a schematic diagram of the structure of the collected image of the curved surface after correction in the present invention.

Fig. 11 is a schematic diagram of the structure of the original plane acquisition image compared with the original plane acquisition image according to the present invention.

FIG. 12 is a schematic diagram of the structure of the corrected curved surface acquisition image compared with the original planar acquisition image according to the present invention.

Detailed ways

Example 1:

The geometric distortion correction method of deep hole inner surface image based on multi-line structured light of the present invention comprises the following steps:

Step 1, model building,

As shown in Figure 1, for common deep hole parts, the 3D MAX software is used to establish its model, that is, the deep hole inner surface model, DIM model; as shown in Figure 2, the corresponding deep hole inner surface expansion model is established, DIPM model, that is, the inner surface of the deep hole inner surface model is developed into a plane relative to the axis;

Step 2, acquire the collected image,

As shown in Figure 3, in the software simulation environment, the structured light laser stripe is projected onto a certain surface of the DIM model to obtain the surface acquisition image; as shown in Figure 4, the same structured light laser stripe is projected onto the same position of the DIPM model The surface at the place, to obtain the plane acquisition image;

Step 3, Gaussian lattice feature pattern,

According to the actual characteristics of the surface geometry of the above model, a special Gaussian structured light lattice pattern is designed, and each Gaussian structured light point is shown in Figure 5;

Step 4, scan the measured object with the characteristic pattern,

The designed Gaussian structured light lattice pattern is projected onto the surface of the same position of the DIM model and the DIPM model respectively, and the collected images are shown in Figure 6 and Figure 7 respectively; the image is preprocessed using the edge detection algorithm based on the gray level difference, as shown in the formula ( As shown in 1), identify all Gaussian points in the image, and obtain the abscissa matrix V _x and ordinate matrix V _y for recording the image position of Gaussian points, as shown in formula (2) and formula (3) respectively; three times The spline difference algorithm interpolates all the Gaussian structured light points in the same row, as shown in Figure 8 and Figure 9 respectively; compares the image positions between the corresponding Gaussian structured light points in the two types of images, and calculates the mutual deformation relationship;

; (1)

; (2)

. (3)

Step five, image geometry correction,

Relying on the image position relationship between the corresponding Gaussian points in the obtained two types of images, the surface acquisition image of the multi-line structured light acquired in the actual simulation process is corrected, and the original surface acquisition image is corrected, as shown in Figure 10; The original plane acquisition image is compared with the original plane acquisition image, as shown in Figure 11; the corrected curved surface acquisition image is compared with the original plane acquisition image, as shown in Figure 12; the horizontal coordinate correction is shown in formula (4), and the vertical Coordinate correction is shown in formula (5):

;(4)

. (5)

Comparing Figure 11 and Figure 12, it can be seen that the slope of the corrected surface acquisition image and the original plane acquisition image are basically the same, and the stripes basically coincide, which lays the foundation for the calculation of the vertical distance between the diagnostic stripes.

The geometric distortion correction method of deep hole inner surface image based on multi-line structured light of the present invention obtains its three-dimensional shape by scanning the deep hole inner surface with multi-line structured light fringe pattern; scans the deep hole inner surface with Gaussian structured light lattice pattern and the corresponding expansion model to determine the deformation law of the inner surface model; in the analysis of the expandable surface model, the cubic spline interpolation algorithm is used to interpolate from the horizontal direction and fit all Gaussians on a single line in each direction Points, and then construct the distortion model of the inner surface of the deep hole, which is convenient for subsequent comparison and correction.

The embodiment of the present invention can be applied to the correction of the geometric distortion existing in the collected images of the inner surface of various common types of deep-hole parts, and the calculation is simple, the precision is high, non-contact scanning is adopted, and the scope of application is wide.

The above-mentioned embodiments are only preferred implementation modes of the present invention, so all equivalent changes or modifications made according to the structures, features and principles described in the scope of the patent application of the present invention are included in the scope of the patent application of the present invention.

Claims (9)

1. A deep hole inner surface image geometric distortion correction method based on multi-line structured light, characterized in that: comprising the following steps: Step 1, establishing a deep hole inner surface model and a corresponding deep hole inner surface expansion model; Step 2, obtaining a curved surface acquisition image for the deep hole inner surface model and a plane acquisition image for the deep hole inner surface unfolded model; Step 3, designing a Gaussian structured light lattice pattern; Step 4, continue to interpolate the Gaussian points in the same row or column; Step 5: Based on the image position and deformation relationship between the corresponding Gaussian structured light points in the curved surface acquisition image obtained by scanning the inner surface of the deep hole with the Gaussian structured light lattice pattern and the plane acquired image, the image distortion is corrected. 2. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 1, characterized in that the specific operation method of the first step is as follows: for common deep hole parts, 3D MAX The software establishes the inner surface model of the deep hole; then establishes the corresponding expansion model of the inner surface of the deep hole. 3. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 2, characterized in that, the deep hole inner surface expansion model in the step 1 is obtained by combining the deep hole inner surface model The inner surface is unfolded into a plane relative to the axis, and the plane model of the inner surface of the deep hole after unfolding is obtained. 4. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 1, characterized in that, the specific operation method of the second step is as follows: using the pre-set multi-line structured light fringe pattern Obliquely projected onto the surface of the measured object, and at the same time use the camera to obliquely shoot the structured light scanning area to obtain the required acquisition image. Specifically, in the software simulation environment, the structured light laser stripes are respectively projected to the same model of the two types of corresponding models. The surface at the position is obtained, and the corresponding acquisition images are obtained respectively. 5. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 1, characterized in that, the specific operation method of the third step is as follows: the structured light points according to the Gaussian distribution are arranged according to the row and column matrix Form arrangement, according to the complexity of the surface geometry of the object to be measured, the density of the light spots after the arrangement is changed. If the surface geometry is complex, the light spot density will increase, otherwise the density will decrease. 6. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 1, characterized in that, the specific operation method of said step 4 is as follows: according to the cubic spline interpolation algorithm, interpolation from the horizontal direction , Fitting all Gaussian points on a single line in each direction, and obtaining the corresponding relationship between the corresponding interpolated fitting curves in the images collected by the two types of models. 7. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 1, characterized in that, the specific operation method of the fifth step is as follows: project the Gaussian structured light lattice pattern into the deep hole respectively. The inner surface model of the hole and the planar model of the inner surface of the deep hole are expanded, and the corresponding curved surface acquisition image and plane acquisition image are obtained respectively; the image position and the corresponding relationship between the Gaussian structured light spots on the respective images are analyzed, and then through interpolation Algorithms construct complete deformation function curves to know the corrections for geometric distortions contained in acquired images of curved surfaces. 8. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 7, characterized in that, in the step 5, the curved surface acquisition image and the plane acquisition image are processed, and the processing method is as follows: Use the edge detection algorithm based on gray level difference to preprocess the image, as shown in the following formula: ; (1) Identify all Gaussian points in the image, and obtain the abscissa matrix Vx and ordinate matrix Vy that record the image position of the Gaussian points, as shown in the following formulas respectively: ; (2) ; (3) The cubic spline difference algorithm is used to interpolate all the Gaussian structured light points in the same row; the image positions between the corresponding Gaussian structured light points in the two types of images are compared, and the mutual deformation relationship is calculated. 9. The method for correcting geometric distortion of deep hole inner surface images based on multi-line structured light according to claim 7, characterized in that, the image geometric correction method in step 5 is as follows: relying on the corresponding Gaussian in the two types of images acquired The image position relationship between the points is corrected for the surface acquisition image of the multi-line structured light acquired in the actual simulation process, and the original surface acquisition image is corrected; the horizontal coordinate correction and the vertical coordinate correction are respectively shown in the following formula: ;(4) (5).