14 results about "Pinhole camera model" patented technology

The invention belongs to the technical field of manipulator tool calibration, and particularly relates to a manipulator micro-pen direct writing tool calibration method, device and system based on machine learning and monocular vision. Comprising the steps that an industrial camera and a direct-writing micro pen are installed under a manipulator tail end coordinate system, the manipulator and the industrial camera are connected to a PC end to complete hand-eye calibration, and a hand-eye matrix is obtained; installing different direct-writing micro-pens, collecting data, and obtaining a plurality of groups of micro-pen central point pixel coordinates and space coordinates in a tail end coordinate system to form point pair data; the hand-eye matrix parameters are optimized through the point pair data; establishing an error compensation model through a pinhole camera model and the single-pixel step length adjustment factor, and carrying out collaborative optimization on the single-pixel step length adjustment factor by adopting point pair data and the optimized hand-eye matrix in combination with the error compensation model; and calculating space coordinates under pixel coordinates by using the optimized hand-eye matrix and parameters to complete calibration. According to the invention, high-efficiency and high-precision calibration can be carried out on different models of manipulators and different specifications of micro-pen printing heads.

The invention discloses a mobile robot formation finite time control method based on visual servo. According to the method, firstly, a relative kinematics model between a navigation robot and a following robot is established under a world coordinate system, the relative kinematics model is converted into a relative vision kinematics model in an image space in combination with a pinhole camera model, and an image error vector is obtained through normalization; for the error model, a non-linear observer is designed to estimate the unknown speed of the piloting robot, a parameter adaptive updating law is provided to process the uncertain feature height, and a finite time controller based on the linear speed and the angular speed is further designed to realize accurate tracking of the piloting robot. The information required by the method can be acquired from the image, the method does not depend on local communication or GPS between the robots, the communication and sensor loads are small, the tracking control precision is good, the system response speed is high, and the method is suitable for efficient cooperative control in the multi-robot formation task.

The application relates to a submarine cable defect detection method and device based on multi-modal perception space correlation, which comprises the following steps: synchronously acquiring submarine cable observation data by using an underwater detection platform, preprocessing the submarine cable observation data, and correlating the submarine cable observation data through a three-dimensional space transformation matrix; adopting a residual network as a backbone network, performing multi-scale fusion through a feature pyramid network structure, and acquiring fusion features; in the multi-scale fusion, a water quality prior parameter is introduced as a gate factor to adjust feature weights, and the water quality prior parameter comprises a light attenuation coefficient of a current water area and a sonar signal-to-noise ratio; the fusion features are input into a pre-trained target detection head, the submarine cable defect category and a bounding box are acquired, laser ranging information is combined, the actual physical size of the defect is calculated according to a pinhole camera model, and submarine cable defect three-dimensional detection results are output, so that submarine cable defect detection is completed. Compared with the prior art, the application realizes automatic identification of submarine cable defects in a complex underwater environment and is suitable for underwater submarine cable defect detection tasks.

The invention discloses a visible light three-dimensional positioning system based on a virtual beacon. The visible light three-dimensional positioning system comprises a signal transmitting end and a positioning terminal, the signal transmitting end encodes and modulates the unique identifier LED-UID of the LED lamp in a world coordinate system set in an indoor scene, and drives the LED lamp to emit UID modulated light by outputting a voltage signal; and the positioning terminal obtains the UID modulation light through a camera on the mobile device, and interprets and calculates the three-dimensional coordinates of the positioning terminal in the world coordinate system so as to obtain the three-dimensional coordinates of the mobile device. According to the system, a high-precision single-lamp three-dimensional positioning scheme is provided on the basis of a pinhole camera model and a perspective projection principle, and height and angle parameters necessary for positioning are solved through combined comparison of a virtual beacon for positioning reference constructed by a simulation modeling platform and actual camera imaging; the positioning terminal can realize accurate three-dimensional positioning without additional auxiliary equipment, and the anti-interference capability of the system is improved.

The invention provides an auricular point position identification and tracking method and device in a three-dimensional space, and belongs to the technical field of intelligent detection, identification and tracking of auricular points in traditional Chinese medicine. The method comprises the following steps: S1, constructing an ear three-dimensional point cloud model based on an obtained human ear image sequence; s2, positioning the three-dimensional space position of each auricular point in the ear three-dimensional point cloud model, and taking the three-dimensional space position of each auricular point as a positioning reference basis of each auricular point under different visual angles; and S3, mapping the three-dimensional space position of each auricular point in the ear three-dimensional point cloud model to a two-dimensional view under the to-be-tracked view angle based on a pinhole camera model according to the camera parameter of any to-be-tracked view angle, so as to realize cross-view-angle auricular point identification and tracking. According to the method, continuous, stable and accurate tracking of the auricular point under multiple view angles can be realized, the method has strong robustness and occlusion recovery capability, and meanwhile, the method does not need to label samples of different angles, so that the labeling cost is reduced, and the efficiency is improved.

A method for constructing an infrared imaging dataset of gas leakage based on computational fluid dynamics includes steps of: establishing a three-dimensional physical model of a gas leakage field scene; meshing and simulating based on the computational fluid dynamics to obtain leaking gas mole fractions of each mesh under time steps, and constituting three-dimensional gas concentration data corresponding to each frame; using optical gas imaging based on a pinhole camera model, imaging the three-dimensional gas concentration data to obtain initial images, and calculating gas concentration path-lengths corresponding to pixel points in each frame of the initial images; and performing maximum-minimum value normalization and generating grayscale images, thereby constructing the infrared imaging dataset of the gas leakage. The method can not only be used in leakage classification tasks and localization tasks, but can also be used in training tasks related to the leakage concentration.

An apparatus and a method for generating a speech synthesis image are provided. The apparatus for generating a speech synthesis image, according to one embodiment, comprises: a head movement prediction module, which estimates three-dimensional geometric transformation of a head movement of a person between a first image and a second image of the person, calculates, on the basis of a pinhole camera model, a plurality of transformation matrices for applying the three-dimensional geometric transformation at a plurality of preset depths, and predicts the head movement of the person between the first image and the second image by using the plurality of transformation matrices; a facial expression movement prediction module for predicting a facial expression movement of the person between the first image and the second image on the basis of a pre-trained facial expression movement prediction model; and a movement generation module for transforming the first image on the basis of the predicted head movement and facial expression movement.

The invention discloses an intelligent ball picking robot visual identification method, and relates to the technical field of robot visual identification and automatic control. Tennis recognition precision is low in a complex light environment. The upper computer converts the RGB image of the tennis area into an HSV image and then processes the HSV image into a black and white binary image; converting the RGB image of the tennis ball area into a grayscale image, dividing the grayscale image into a background image and an image containing a tennis ball, and detecting all round object contours from the image containing the tennis ball by adopting a Hough circle detection method; fusing the processed black and white binary image and all circular object contours to obtain an image containing all circular objects, and selecting a circular object image of which the radius pixel accords with a preset radius pixel from the image as a candidate target; the pinhole camera model adopts a similar triangle principle to obtain the distance from a small hole of the pinhole camera model to each actual tennis ball; the pinhole camera model obtains an included angle between each actual tennis ball and the horizontal plane by adopting a lens transformation principle; therefore, tennis ball position identification is realized. And identifying tennis balls.

The invention discloses a method, a system and equipment for detecting personnel in a forbidden area under a cantilever and a storage medium, and relates to the technical field of image processing and safety monitoring, and the method comprises the steps: obtaining a video image collected by a monocular camera, recognizing a preset area in the video image through a target detection model, and outputting a bounding box corresponding to the preset area; based on a pinhole camera model and camera internal parameters, performing monocular back projection on pixel points in the bounding box, and mapping the pixel points to a ground coordinate system to obtain three-dimensional space position information; and constructing a forbidden zone based on the three-dimensional space position information, identifying whether the person enters the forbidden zone illegally by judging whether the three-dimensional projection point of the person falls into the forbidden zone, and outputting an alarm signal. According to the method, automatic extraction of key targets in a working site is realized, three-dimensional position information of the targets is acquired under the condition of no depth sensor assistance, and real-time identification and early warning response of illegal intrusion behaviors in a working area are realized.

The invention discloses an image deblurring method based on an IMU (Inertial Measurement Unit) enhanced double-branch codec network. The method comprises the following steps: synchronously acquiring an RGB (Red, Green, Blue) blurred image and IMU original data in a global blurred scene; preprocessing the IMU original data to obtain camera motion parameters between adjacent image frames; generating an IMU (Inertial Measurement Unit) expected optical flow field of an image plane based on a pinhole camera model; and acquiring a trained dual-branch codec network model based on IMU enhancement, and inputting the RGB blurred image and the expected optical flow field as the trained dual-branch codec network model based on IMU enhancement to obtain a deblurred image and a pixel-level motion prediction field. According to the method, image deblurring can be efficiently achieved, the method is further used for SLAM, and the problem of SLAM performance degradation in a global blurred scene is solved.

This invention discloses a method for detecting fires in outdoor oil-immersed transformers based on binocular 3D vision. Based on the imaging effect of an infrared thermal imager, the geometric imaging model of the infrared thermal imager is treated as a pinhole camera model, and a pinhole camera model is established. A checkerboard calibration plate is fabricated. Using external transmission infrared radiation, the specially made checkerboard calibration plate is attached to the surface of the heat source. Multiple infrared images of the calibration plate from different angles are acquired, and then calibration is performed using a camera calibration tool provided by OpenCV. The intersection center of the horizontal and vertical bars of a ladder is used as a feature reference point for calibrating the extrinsic parameters of the infrared thermal imager. The EPnP algorithm and ICP-based 3D-3D pose estimation are used to calculate the camera pose. After registration of the infrared and visible light images, abnormal high-temperature areas and target points are extracted. This invention solves the problem of large errors in single-band infrared fire detection methods due to interference from outdoor sunlight, humidity, and wind speed, and improves the accuracy of calibration.

A visual identification method for an intelligent ball picking robot comprises the steps that 1, a visual identification system is adopted to collect RGB images of a tennis area in real time, and the RGB images are transmitted to an upper computer; step 2, color feature recognition; shape feature recognition; step 3, fusing the processed black and white binary image and all circular object contours to obtain an image containing all circular objects, and selecting a circular object image of which the radius pixel accords with a preset radius pixel from the image as a candidate target; 4, the pinhole camera model obtains the distance from a small hole of the pinhole camera model to each actual tennis ball and the included angle between each actual tennis ball and the horizontal plane according to the pixel size of the candidate target, the size of the actual tennis ball corresponding to the candidate target and the focal length of the pinhole camera model. According to the invention, all half-shielded and half-recognized tennis images can be recognized, tennis balls which are not easy to recognize due to light influence can be accurately recognized, missing detection is avoided, the robustness is high, and false detection is not easy to occur.

This disclosure relates to a method, apparatus, device, and medium for distortion correction of images from railway communication base station inspections. The method includes: extracting the true distorted pixel coordinates from images of marker points captured by a camera at multiple preset locations; determining the first calibration pixel coordinates of the marker points based on a pinhole camera model; transforming the first calibration pixel coordinates to a normalized plane to obtain first calibration plane coordinates; performing a forward distortion transformation on the first calibration plane coordinates and then performing a pixel coordinate restoration transformation to obtain the expected distorted pixel coordinates; calculating the target deviation between the expected distorted pixel coordinates and the true distorted pixel coordinates at each preset location; using the minimization of the target deviation as the optimization objective, iteratively updating the distortion parameters using a nonlinear optimization algorithm until a preset convergence condition is met to obtain the target distortion parameters. This disclosure can improve the accuracy of distortion parameters, thereby improving the accuracy of image distortion correction.

The present application relates to the technical field of camera calibration, and discloses a camera calibration method and system, a device, a storage medium, and a program product. The method comprises: on the basis of correspondences between spatial points and image points of a single image, constructing a linear equation about a rotation matrix, a translation vector, and a focal length element on the basis of a pinhole camera model; converting the linear equation into a homogeneous system of equations about a rotation matrix element and a focal length; using a null-space analysis method and a rotation matrix orthogonality constraint to calculate the homogeneous system of equations so as to obtain the rotation matrix element and the focal length, and selecting initial values for subsequent iterative optimization; and constructing a cost equation on the basis of the rotation matrix orthogonality constraint, and on the basis of the initial values, performing iterative optimization to obtain a global optimal solution for camera calibration. In the present application, a constraint that is not based on intermediate variables is constructed on the basis of the pinhole camera model, and a null-space analysis framework based on a rotation matrix constraint is used to solve for related parameters, thereby shortening operation time and improving the robustness and stability of point distribution.