Infrared vision positioning method for concrete pumping liquid level in tube of steel tubular arch bridge

An infrared vision positioning method for a concrete pumping liquid level in a tube of a steel tubular arch bridge, comprising the following steps:

• step 1:
• using a calibration plate to calibrate an infrared thermal imaging camera, and obtaining an intrinsic matrix K and a distortion parameter D of the infrared thermal imaging camera;
• step 2:
• setting up the infrared thermal imaging camera and adjusting an orientation of the infrared thermal imaging camera, wherein a half of the steel tubular arch bridge from a pumping port to a top of an arch is comprised in an infrared thermal imaging picture before a concrete pumping construction, and adjusting a focal length and related parameters to make the infrared thermal imaging picture clear and stable;
• step 3:
• turning on the infrared thermal imaging camera to collect an image after starting a pumping of a concrete filled steel tubular, due to a slow rise of a liquid level, collecting a frame image at intervals of 1 min, and performing a distortion correction on collected images;
• step 4:
• importing a coordinate of a feature point, and solving a camera pose at a first current time by a perspective decoupling (PNP) algorithm to obtain a camera pose matrix [R|T], and transforming a world coordinate of the feature point into a camera coordinate system Oc;
• step 5:
• obtaining a matrix [C|t] by inverting a [R|T] matrix, constructing a pose transformation matrix [W1] according to a rotation matrix C and performing a rotation transformation on the camera coordinate system Oc to obtain a camera coordinate system
• O
• c
• ′
• ,
• constructing a pose transformation matrix [W2] by a translation vector t at a second current time, and translating the camera coordinate system
• O
• c
• ′
• to a front of a world coordinate system by a translation transformation of the camera coordinate system
• O
• c
• ′
• ;
• step 6:
• projecting the feature point onto a camera plane by an orthographic projection to obtain a 2D orthographic projection coordinate of the feature point;
• step 7:
• transforming a pixel coordinate of a feature point under a default pixel coordinate system Op into a pixel coordinate system
• O
• p
• ′
• with the world coordinate system as an origin;
• step 8:
• obtaining a perspective transformation matrix P by a one-to-one correspondence between the 2D orthographic projection coordinate and a 2D pixel coordinate in the pixel coordinate system
• O
• p
• ′
• ;
• step 9:
• performing a perspective transformation on an arch bridge image of a current frame to obtain an orthographic projection arch bridge image;
• step 10:
• calculating an image pixel scale S=L/d;
• step 11:
• detecting a liquid level position on the orthographic projection arch bridge image by a target detection algorithm, and returning to default pixel coordinates of the liquid level;
• and
• step 12:
• transforming the default pixel coordinates of the liquid level into a real coordinate system of the liquid level with an arch foot of the steel tubular arch bridge as the origin, calculating a mileage x and an elevation y of the liquid level at a third current time by the image pixel scale to complete positioning.