CN109215024B

CN109215024B - Method for automatically testing defective black light lamp by android

Info

Publication number: CN109215024B
Application number: CN201811105747.8A
Authority: CN
Inventors: 杨芸; 李龙; 李婷婷; 李秀芬; 陈翠丽; 李明
Original assignee: Shanghai Chindt Systems And Services Co ltd; Donghua University
Current assignee: Shanghai Chindt Systems And Services Co ltd; Donghua University
Priority date: 2018-09-21
Filing date: 2018-09-21
Publication date: 2021-07-20
Anticipated expiration: 2038-09-21
Also published as: CN109215024A

Abstract

The invention discloses a method for automatically testing a defective black light lamp by an android. The method comprises the following specific operation steps of shooting a nondestructive inspection workpiece through a USB high-definition microspur camera, and automatically calculating the length and the size of a crack at two points of the crack after amplifying or reducing a picture, wherein the specific operation steps are as follows: (1) based on a camera model, mapping from one point M in space to an image plane point M is converted through a projection matrix P, and the coordinate sm of the image point is K [ R | T ] M; (2) the space object points are projected to the left image and the right image through the two projection matrixes P, the coordinates of the two projection matrixes P and the coordinates of the two image points on the images are obtained according to the theory of three-dimensional reconstruction, and the three-dimensional coordinates of the target point are positioned by using a triangulation method, so that the length and the size of the crack are calculated. The invention has the beneficial effects that: the crack length and the size are automatically calculated, so that an operator can directly obtain bad work information on a screen of the handheld device, and production efficiency is improved.

Description

Method for automatically testing defective black light lamp by android

Technical Field

The invention relates to the technical field of black light lamp correlation, in particular to a method for automatically testing a defective black light lamp by an android system.

Background

The black light lamp is a specially-made gas discharge lamp, and emits 330-400nm ultraviolet light waves which are light insensitive to human beings, so the lamp made of the ultraviolet light insensitive to human beings is called a black light lamp. Black light lamps look like ordinary fluorescent or incandescent bulbs, but they are in some places completely different. It is used by maintenance personnel to find hidden cracks in the machine-a small amount of fluorescent dye is injected into the fuel supply and then the device is illuminated with a black light lamp. For example, a fluorescent dye is added to a refrigerant of an air conditioner, so that a leak invisible to the naked eye of the air conditioner can be detected. However, the original nondestructive inspection instrument cannot automatically calculate the length of the crack, and the length and the size of the crack need to be manually measured by a measuring tool.

Disclosure of Invention

The invention provides a method for automatically testing a defective black light lamp by an android system, which can automatically calculate the length of a crack, so as to overcome the defects in the prior art.

In order to achieve the purpose, the invention adopts the following technical scheme:

a method for automatically testing a defect black light lamp by an android system comprises the following specific operation steps of shooting a nondestructive flaw detection workpiece through a USB high-definition microspur camera, and automatically calculating the length and the size of a crack by two points of the crack after amplifying or reducing a picture:

(1) based on a camera model, mapping from a point M in space to an image plane point M is converted through a projection matrix P, and coordinates sm of the image point are obtained as K [ R | T ] M, wherein K is a camera parameter matrix, R, T is a rotation and translation matrix of a camera coordinate system relative to a world coordinate system, and sm is a homogeneous coordinate of the image point;

(2) the space object points are projected to the left image and the right image through the two projection matrixes P, the coordinates of the two projection matrixes P and the coordinates of the two image points on the images are obtained according to the theory of three-dimensional reconstruction, and the three-dimensional coordinates of the target point are positioned by using a triangulation method, so that the length and the size of the crack are calculated.

Through a USB high-definition microspur camera, a nondestructive inspection workpiece is shot, after a picture is enlarged or reduced, the length and the size of the crack are automatically calculated by two points of the point crack, so that an operator can directly obtain poor working information on a screen of a handheld device, and the production efficiency is improved.

Preferably, in step (1), a distance-fixing method is adopted, 10 pictures are taken by a camera, and a camera parameter matrix K is obtained by using a calibration program implemented by a Zhang friend calibration method algorithm.

Preferably, in step (1), taking the camera coordinate system for shooting the first picture as a reference coordinate system, R of the first camera position is the identity matrix I, and T is not translated to zero; and (2) applying an antipodal geometric theory, solving the R and T at the second position by utilizing the corresponding relation of two pictures, finding out more than 7 pairs of matching points, calculating the intrinsic relation of the two pictures, calculating a basic matrix F by using the matched characteristic point pairs, then calculating an essential matrix E, wherein the essential matrix E is equal to a transposition matrix of a camera parameter matrix K, multiplied by the basic matrix F, multiplied by the camera parameter matrix K, carrying out SVD decomposition on the essential matrix E, and the obtained final column of the U matrix is multiplied by a single value to obtain a translation matrix T, and the U matrix is multiplied by an antisymmetric matrix, multiplied by the V transposition matrix to obtain a rotation matrix R.

Preferably, one or more of 7-point method, 8-point method and RANSAC are selected according to the number of matching points to calculate the fundamental matrix F.

The invention has the beneficial effects that: the crack length and the size are automatically calculated, so that an operator can directly obtain bad work information on a screen of the handheld device, and production efficiency is improved.

Detailed Description

The invention is further described with reference to specific embodiments.

(1) based on a camera model, mapping from a point M in space to an image plane point M is converted through a projection matrix P, and coordinates sm of the image point are obtained as K [ R | T ] M, wherein K is a camera parameter matrix, R, T is a rotation and translation matrix of a camera coordinate system relative to a world coordinate system, and sm is a homogeneous coordinate of the image point; taking 10 pictures by a camera in a distance-setting mode, and obtaining a camera parameter matrix K by using a calibration program realized by a Zhang Zhengyou calibration method algorithm; taking a camera coordinate system for shooting a first picture as a reference coordinate system, wherein R of the position of the first camera is an identity matrix I, and T is zero without translation; by using the theory of epipolar geometry, R and T at the second position are solved by utilizing the corresponding relation of two images, the internal relation of the two images is calculated after more than 7 pairs of matching points are found, a basic matrix F is calculated by using matched characteristic point pairs, and one or more of a 7-point method, an 8-point method and RANSAC is/are selected to be used for calculating the basic matrix F according to the number of the matching points; then, an essential matrix E is calculated, the essential matrix E is equal to a transposed matrix of a camera parameter matrix K, the basic matrix F is multiplied by the essential matrix E, the camera parameter matrix K is multiplied by the essential matrix E, SVD decomposition is carried out on the essential matrix E, the translation matrix T is obtained by multiplying a single value by the last column of the obtained U matrix, and the rotation matrix R is obtained by multiplying the inverse symmetric matrix by the U matrix and then multiplying the V transposed matrix;

According to the composition of the projection matrix P, the key to the whole system is to obtain the camera parameter matrix K and to obtain the rotation and translation (R | T) matrix which is the rigid body transformation experienced by the camera when the second image is shot. The method is easy to operate, and because the method is distance, about 10 pictures are shot by a mobile phone Camera, and a Camera parameter matrix K can be obtained by using a Calibration program realized by an algorithm of the method. The implementation on the Android handheld device is that the essential matrix E can be calculated only by selecting seven points, seven colored dots can be seen to be arranged in a Beidou seven-star shape after entering a point selection interface, and the dots are moved to positions with obvious characteristics, such as corner points, through touch. In order to realize real-time processing, the core operation code is called to OpenCv (open source computer visual library), and after the C + + program is compiled, the source code is compiled into a dynamic link library which can be called by the Android-end java program through NDK.

Claims

1. A method for automatically testing a defect black light lamp by an android is characterized in that a nondestructive flaw detection workpiece is shot by a USB high-definition microspur camera, the length and the size of a crack are automatically calculated by two points of the crack after the picture is enlarged or reduced, and the specific operation steps are as follows:

(1) based on a camera model, mapping from a point M in space to an image plane point M is converted through a projection matrix P to obtain the coordinate sm = K [ R | T ] M of the image point, wherein K is a camera parameter matrix, R, T is a rotation and translation matrix of a camera coordinate system relative to a world coordinate system, and sm is the homogeneous coordinate of the image point; taking 10 pictures by a camera in a distance-setting mode, and obtaining a camera parameter matrix K by using a calibration program realized by a Zhangyingyou calibration method algorithm; taking a camera coordinate system for shooting a first picture as a reference coordinate system, wherein R of the position of the first camera is an identity matrix I, and T is zero without translation; by using the theory of antipodal geometry, R and T at the second position are solved by utilizing the corresponding relation of two pictures, the intrinsic relation of the two pictures is calculated after more than 7 pairs of matching points are found, a basic matrix F is calculated by using matched characteristic point pairs, then an essential matrix E is calculated, the essential matrix E is equal to a transposed matrix of a camera parameter matrix K which is multiplied by the basic matrix F and then multiplied by the camera parameter matrix K, SVD decomposition is carried out on the essential matrix E, the last column of the obtained U matrix is multiplied by a single value to obtain a translation matrix T, and the U matrix is multiplied by an antisymmetric matrix and then multiplied by a V transposed matrix to obtain a rotation matrix R;

2. The method of claim 1, wherein the fundamental matrix F is calculated by selecting one or more of 7-point method, 8-point method and RANSAC according to the number of matching points.