CN108769459A - Multiple spot laser Full-automatic oblique angle shot based on image procossing corrects system - Google Patents

Abstract

In the process of automatic detection using image processing methods, the multi-point laser automatic correction method based on coordinates and angles can quickly and automatically correct images taken at oblique angles, and ensure the correction accuracy. 1. Coordinate-based multi-point laser automatic correction method: scheme one of the present invention utilizes 4 distances measured and 4 dotting coordinate information, by the method of spatial three-dimensional geometry method and similarity theorem, deduces the dotting coordinate after correction, Then calculate the correction matrix and correct the image according to the perspective transformation method, so as to ensure the precision requirement of correction. 2. Angle-based multi-point laser automatic correction method: the present invention utilizes the measured 4 distances and the calibrated 4 dot angle information, through the method of spatial three-dimensional geometry, the conversion of the spherical coordinate system and the rectangular coordinate system, and the method of similarity theorem , deduce the dot coordinates before and after correction, and then calculate the correction matrix and correct the image according to the perspective transformation method, which ensures the accuracy requirements of the correction.

Description

Multi-point laser automatic oblique angle shooting correction system based on image processing

technical field

The invention relates to a multi-point laser automatic correction system for detection of building walls (such as patterns, cracks, etc.), in particular to a camera with a laser range finder to take pictures of inclined walls and perform rapid The technology of correction belongs to the fields of computer vision and image processing.

Background technique

Due to factors such as the camera shooting angle and the tilt of the building, the captured image may be distorted and distorted, and the distorted and distorted image can be corrected by using image correction technology. Image correction refers to the restorative processing of distorted images. As an image processing technology, it continues to develop. At present, image correction technology plays an important role in many fields such as education, military affairs, medicine, aerospace, and construction engineering.

At present, the correction of the captured distorted image mainly adopts the method of manual labeling, extracts the edge of the distorted image, and then calculates the coordinates of the marked point in the corrected image according to the coordinates of the marked distorted image and calculates the correction matrix, and finally uses the correction matrix to correct the image. Correction.

Current image correction techniques usually have the following problems:

(1) The manual labeling method is mainly used for the captured photos, which is time-consuming and labor-intensive, and the efficiency is low.

(2) Neglecting the distortion at the end of the image near the lens, it lacks accuracy.

(3) The lack of accurate distance data also affects the correctness of subsequent automatic image stitching.

The present invention adopts a multi-point laser full-automatic correction method, without manual marking, to quickly and automatically correct tilted wall photos, and to accurately correct distances, thereby enabling real-time automatic splicing of scanned pictures, providing large-area Automatic and rapid detection of building walls (such as patterns, cracks, etc.) provides conditions.

Contents of the invention

The purpose of the present invention is to solve the problems of time-consuming and labor-intensive manual labeling and lack of accuracy of the correction matrix in the current image correction process, and propose two multi-point laser automatic correction methods based on coordinates and angles, including: 1. Using laser distance measurement 2. Considering that the traditional correction method ignores the distortion factor close to the lens end, the correction method of the present invention calculates the correction matrix more accurately, and can quickly and automatically correct the tilted wall photos, improving the correction accuracy.

In order to solve the above problems, the present invention proposes two technical solutions:

Solution 1: Multi-laser automatic correction method based on coordinates

A multi-point laser automatic correction system for detection of building walls (such as patterns, cracks, etc.), the hardware includes a CCD camera and 4 single-point laser range finders. Install 4 laser rangefinders at the same position of the CCD camera at 4 different angles, use the laser rangefinder to measure 4 distances at 4 angles at the same time, and take photos to record the dots while the laser rangefinder is marking the distance The position is calculated and corrected according to the laser marking position (x1, y1) ~ (x4, y4) in the photo. Solution 1 of the present invention utilizes the measured 4 distances and 4 dotted coordinate information to derive the corrected dotted coordinates through the method of spatial three-dimensional geometry and similarity theorem, then calculates the correction matrix according to the perspective transformation method, and finally uses the corrected The matrix corrects the image to ensure the accuracy requirements of the automatic correction system.

Solution 2: Multi-laser automatic correction method based on angle

A multi-point laser automatic correction system for detection of building walls (such as patterns, cracks, etc.), the hardware includes a CCD camera and 4 single-point laser range finders. Install the 4 laser rangefinders at the same position of the CCD camera, and use the laser rangefinder to measure the 4 distances at the same time at the 4 angles calibrated in advance. α2,β2)(α3,β3)(α4,β4) calculates the correction. The present invention utilizes the measured 4 distances and the calibrated 4 dot angle information to deduce the dot coordinates before and after correction through the method of spatial three-dimensional geometry, the conversion of the spherical coordinate system and the rectangular coordinate system, and the similarity theorem, and then according to the perspective The transformation method is used to calculate the correction matrix, and finally the image is corrected by using the correction matrix, which ensures the precision requirement of the automatic correction system.

Beneficial effects of the present invention: when photographing the walls of buildings, the multi-point laser full-automatic calibration method is used to correct the obliquely photographed images. Considering from the economic level, manual marking is not required, which can greatly reduce manpower and improve work efficiency; considering from the technical level , It can quickly and automatically correct the tilted wall photos, and can accurately correct the distance to improve the calibration accuracy.

Description of drawings

Figure 1 Flowchart of coordinate-based multi-laser automatic calibration method

Figure 2 Oblique shot

Figure 3 Coordinate-based multi-laser dot diagram

Figure 4 Coordinate system diagram of multi-laser dot marking space based on coordinates

Figure 5 Experiment 1 Coordinate-based multi-laser automatic calibration results

Figure 6 Experiment 2 Coordinate-based multi-laser automatic calibration results

Figure 7 Experiment 1 Comparison between the vertical shooting picture and the calibration picture

Figure 8 Comparison of the vertical shooting image and the calibration image in experiment 2

Fig.9 Scene with cracked wall - the result of multi-laser automatic correction based on coordinates

Figure 10 Flow chart of angle-based multi-laser automatic correction method

Figure 11 Spherical coordinate system diagram

Figure 12 Angle-based multi-laser dot diagram

Figure 13 The spatial coordinate system diagram of multi-laser dot marking based on angle

Figure 14 Experiment 3 The results of multi-laser automatic correction based on angle

Figure 15 Experiment 4: Angle-based multi-laser automatic correction results

Figure 16 The comparison between the three vertical shooting pictures and the correction pictures in the experiment

Figure 17 The comparison between the vertical shooting picture and the calibration picture of Experiment 4

Figure 18 The results of multi-laser automatic correction based on angle in scene 2 of the wall with cracks

Detailed ways

Embodiment one

Multi-laser automatic correction method based on coordinates

This proposal elaborates in detail the specific implementation of a coordinate-based multi-laser automatic correction method of the present invention.

Install 4 laser rangefinders at the same position of the CCD camera at 4 different angles, use the laser rangefinder to measure 4 distances at 4 angles at the same time, and take photos to record while the laser rangefinder is marking the distance The dot position is calculated and corrected according to the laser dot position (x1, y1) ~ (x4, y4) in the photo. The correction method involves three calculation methods: spatial solid geometry method, similarity theorem, and perspective transformation.

Based on the measured 4 distances and 4 dot coordinate information: use the spatial three-dimensional geometry method to establish a space Cartesian coordinate system, calculate the plane equation of the wall and the tilt angle of the shooting; use the similarity theorem to calculate the corrected dot coordinates; use The perspective transformation method calculates the correction matrix, and finally uses the correction matrix to correct the image. The flow chart of Scheme 1 is shown in Figure 1.

Specific steps:

In the process of automatic detection of building walls (such as patterns, cracks, etc.) using image processing methods, when using a camera to take pictures of the wall, there must be a situation where the angle of view is tilted as shown in Figure 2.

Step 1: Use the spatial three-dimensional geometry method to establish a spatial rectangular coordinate system, calculate the plane equation of the wall and the tilt angle of the shooting

In Figure 3, (x1, y1) ~ (x4, y4) are the positions of four laser dots, and the four line segments measured by the laser rangefinder are l1, l2, l3, l4; the four distances are L1, L2, L3, L4; establish a space Cartesian coordinate system with the camera coordinates as the origin, as shown in Figure 4.

The coordinates of the four dot positions in the space coordinate system are (x1, y1, z1), (x2, y2, z2), (x3, y3, z3), (x4, y4, z4).

Knowing L1～L4 and, (x1, y1)～(x4, y4) to calculate z1, z2, z3, z4, using the triangle Pythagorean theorem

Then there are:

Because the four dotted coordinates are on the same plane, set the plane equation as Ax+By+Cz+D=0, and substitute (x1,y1,z1)~(x4,y4,z4) into the plane equation respectively, then:

at this time

Calculate the coefficients A, B, C, and D through matrix operations.

Let the optical axis of the camera intersect the plane at point P, set the optical center (0,0,P), and substitute the coordinates of the optical center into the plane equation, CP+D=0, so

Now take the optical center as the benchmark, rotate the plane around the optical center by an angle of θ to make it perpendicular to the optical axis, then the plane equation is:

Calculating the rotation angle is the shooting tilt angle: the distance from the camera to the plane equation Ax+By+Cz+D=0 is

camera to plane The distance of the equation is:

so

so

Step 2: Use the similarity theorem to calculate the corrected dot coordinates

Calculate the laser in the plane At the dotted position, project the 4 laser lines onto the X0Z plane respectively, and the equations of the lines are

Will Substitute into the four straight line equations, l1, l2, l3, l4 and the plane The points of intersection are:

Calculated by the similarity theorem (x1, y1, z1), (x2, y2, z2), (x3, y3, z3), (x4, y4, z4) in the plane Corresponding coordinates on (x1', y1', z1'), (x2', y2', z2'), (x3', y3', z3'), (x4', y4', z4')

so in the same way

Then the four dotting positions of the laser on the plane after rotation are

So far, 8 position coordinates before and after correction are obtained.

Step 3: Calculate the correction matrix using the perspective transformation method, and finally correct the image using the correction matrix

Based on the above, the original dot position and the corrected dot position have a total of 8 coordinates. According to the original dot position, the pixel coordinates (p1, q1), (p2, q2), (p3, q3), (p4, q4), calculate the pixel coordinates corresponding to the dot position after correction as (p1', q1'), (p2', q2'), (p3', q3'), (p4', q4'), a two-dimensional plane passes Perspective transformation, becoming another flat image, this process is defined as:

Among them (pi', qi') are transformed pixel coordinates, (pi, qi) are original pixel coordinates (Note: (i=1,2,3,4)), h1,h2,h3,h4,h5, h6, h7, and h8 are transformation coefficients. For a given set of 4 transformation points, the defined transformation matrix is as follows:

Therefore, given 4 points and points in the corresponding transformation space, as long as any three of them are not in a straight line, 8 transformation parameters can be solved. Thus, the transformation relationship before and after the perspective transformation can be obtained, and the purpose of correcting the image can be achieved by the perspective transformation of the figure; Fig. 5 and Fig. 6 are the results of the checkerboard correction based on the multi-laser automatic correction method based on coordinates; Fig. 7 and Fig. 8 are respectively The comparison between the corresponding vertical shooting picture (the shooting distance is the distance from the camera to the optical center) and the correction picture shows that the size of the checkerboard is basically the same; Figure 9 is the result picture of the correction of the cracked wall in scene one. Based on the experimental results of Example 1, the multi-laser automatic correction method based on coordinates has a good correction effect on obliquely taken photos, and the correction accuracy is higher compared with the original image.

Embodiment two

Multi-laser automatic correction method based on angle

This solution describes in detail the specific implementation of an angle-based multi-laser automatic correction method of the present invention.

Install 4 laser range finders at the same position of the CCD camera, use the laser range finder to measure 4 distances at the same time at the 4 previously calibrated angles, according to the previously calibrated angles of the 4 laser dots (α1, β1) ,(α2,β2),(α3,β3),(α4,β4) calculate the correction. The correction method involves four calculation methods: the three-dimensional geometry method, the conversion between the spherical coordinate system and the rectangular coordinate system, similarity theorem, and perspective transformation.

Based on the measured 4 distances and 4 angles information: use the spatial three-dimensional geometry method, the conversion of the spherical coordinate system and the rectangular coordinate system to establish a spatial rectangular coordinate system, calculate the plane equation of the wall, the tilt angle of the shooting and the initial point. position; use the similarity theorem to calculate the corrected dot coordinates; use the perspective transformation method to calculate the correction matrix, and finally use the correction matrix to correct the image. The flow chart of scheme two is shown in Figure 10.

Specific steps:

In the process of automatic detection of building walls (such as patterns, cracks, etc.) using image processing methods, when using a camera to take pictures of the wall, there must be a situation where the angle of view is tilted as shown in Figure 2.

Step 1: Use the spatial three-dimensional geometry method, the conversion method of the spherical coordinate system and the rectangular coordinate system to establish a spatial rectangular coordinate system, and calculate the plane equation of the wall, the inclination angle of the shooting and the initial point location.

The conversion relationship between the spherical coordinate system and the space Cartesian coordinates is shown in Figure 11, where:

(Note: Assuming that P(x,y,z) is a point in the space, the point P can also use such three ordered numbers To determine, where r is the distance between the origin O and point P; θ is the angle between the directed line segment OP and the positive direction of the z-axis; It is the angle turned from the x-axis to OM in the counterclockwise direction from the positive z-axis, where M is the projection of point P on the XOY plane. )

The conversion of each space Cartesian coordinate in the spherical coordinate system corresponds to two angles, angle θ, Calibrate in advance, 4 dotting positions need to calibrate 4 angles in total; as shown in Figure 12, the angles of 4 laser dots are calibrated in advance as (α1, β1), (α2, β2), (α3, β3), (α4, β4), Let the coordinates A, B, C, and D of the 4 dot positions be (x1, y1, z1), (x2, y2, z2). (x3, y3, z3), (x4, y4, z4), as shown in the figure 13 Establish a space rectangular coordinate system with the camera as the origin, and the intersection point of the optical axis and the object plane is P.

It is known that the distance from the origin measured by the laser rangefinder to A is L1, and according to the conversion relationship between the spherical coordinate system and the rectangular coordinates, there is

in the same way

Among them, L1, L2, L3, L4, α1, α2, α3, α4, β1, β2, β3, β4 are known.

Points A, B, C, and D are on the same plane, and the plane equation is set as Ax+By+Cz+D=0, and (x1, y1, z1) ~ (x4, y4, z4) are respectively substituted into the plane equation, then:

so

Calculate the coefficients A, B, C, and D through matrix operations to get the plane equation where the wall is located.

Seek rotation angle and correction matrix with embodiment one coordinate-based multi-point laser automatic correction method step 2, step 3, Fig. 14, Fig. 15 is the result figure of checkerboard correction to the checkerboard correction method based on the multi-laser automatic correction method of angle; Fig. 16, Fig. Figure 17 is the comparison between the corresponding vertical shooting picture (the shooting distance is the distance from the camera to the optical center) and the correction picture. By comparison, the size of the checkerboard is basically the same; Figure 18 is the result of the correction of the wall with cracks taken in scene two . Based on the experimental results of Example 2, the angle-based multi-laser automatic correction method has a good correction effect on obliquely taken photos, and the correction accuracy is higher compared with the original image.

The above is only an embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the description of the present invention and the contents of the accompanying drawings, or directly or indirectly used in other related technologies fields, all of which are equally included in the scope of patent protection of the present invention.

Claims (3)

1. a kind of multiple spot laser automatic Calibration System of cracking of wall surface detection, it is characterized in that：Using based on coordinate and based on angle Two kinds of multiple spot laser auto-correction methods of degree carry out fast automatic correction to inclined metope photo, and can adjust the distance It is accurately corrected, real-time automatic Mosaic is carried out to scanned picture so as to realize, for splitting for large area building surface The automatic quickly detection offer condition of seam.

2. more laser auto-correction methods according to claim 1 based on coordinate, it is characterized in that：A kind of building wall The multiple spot laser Full-automatic correction system of (such as with pattern, with crack etc.) detection, hardware include CCD camera and 4 single-points Laser range finder.4 laser range finders are mounted on to 4 different angles of CCD camera same position, it is same with laser range finder When got ready in 4 angles and measure 4 distances, while laser range finder gets ranging ready, shooting photo record gets position, root ready Correction is calculated according to laser dotting position (x1, y1)~(x4, y4) in photo.The solution of the present invention one utilizes 4 distances measured Coordinate information is got ready with 4, by the method for space multistory geometric method and correspondence theorem, is derived and is got coordinate ready after correction, Correction matrix is calculated further according to perspective transform method, finally utilizes correction matrix to image rectification, ensure that full-automatic correction system The required precision of system.

3. more laser auto-correction methods according to claim 1 based on angle, it is characterized in that：A kind of building wall The multiple spot laser Full-automatic correction system of (such as with pattern, with crack etc.) detection, hardware include CCD camera and 4 single-points Laser range finder.4 laser range finders are mounted on to the same position of CCD camera, with laser range finder in 4 demarcated in advance A angle is got ready simultaneously measures 4 distances, according to the angle (α 1, β 1) of 4 laser dottings of prior calibration, (α 2, β 2), (α 3, β 3), (α 4, β 4) calculates correction.The present invention gets angle information ready using 4 of 4 distances and calibration that measure, vertical by space Body geometric method, the conversion of spherical coordinate system and rectangular coordinate system, correspondence theorem method, derive correction it is front and back get coordinate ready, Correction matrix is calculated further according to perspective transform method, finally utilizes correction matrix to image rectification, ensure that full-automatic correction system The required precision of system.