224 results about "Monocular vision" patented technology

The invention belongs to the technical field of robot positioning and three-dimensional mapping, and discloses a high-precision map reconstruction method and system based on monocular vision, a medium and equipment. Scene image data are acquired through a monocular image acquisition module, after feature extraction and matching are performed on each frame of image, attitude information acquired by an inertial measurement unit is fused, and pose calculation of a robot is completed through a sparse vision SLAM system; meanwhile, an image dense depth map is generated by a monocular depth estimation model based on an attention mechanism, and the image dense depth map is converted into a single-frame color dense point cloud in combination with image RGB color information. According to the invention, a loose coupling fusion strategy is adopted to carry out spatial registration on a single-frame colored dense point cloud and a robot pose at a corresponding moment, multi-frame data fusion is completed through point cloud splicing and optimization, and a globally consistent three-dimensional dense point cloud map is constructed to realize scene modeling. The method has the characteristics of high robustness and high reconstruction precision, and can be effectively applied to three-dimensional map construction in an outdoor complex environment.

The invention discloses a material movement behavior monitoring method and system based on monocular vision, and relates to the field of material movement monitoring, and the method comprises the steps: firstly collecting a material movement video through a monocular high-speed camera, and extracting a continuous state sequence of a material on a two-dimensional image plane through employing an instance segmentation and time sequence tracking technology; the method is characterized in that a time sequence deep learning model is introduced to analyze a state sequence, and a key event that a material collides with a working surface of equipment is automatically identified, so that the collision moment and contact point coordinates are accurately obtained. In this way, the collision contact point is used as a geometric anchor point for three-dimensional space back calculation, a two-dimensional image track is reversely projected and reconstructed into a three-dimensional space motion track in combination with internal parameters of a camera and physical parameters of an equipment reference plane, then key kinetic parameters such as the speed and the recovery coefficient are calculated, and low-cost and high-precision automatic online monitoring is achieved.

The invention relates to a multi-modal perception fusion target following method and system. The method comprises the following steps: adopting an improved KCF algorithm to realize a closed-loop process of target tracking, multi-scale space construction, feature fusion, response calculation, optimal scale decision and model updating; designing a four-level shielding processing mechanism fusing motion prediction and depth verification, and realizing tracking recovery in a shielding scene through shielding judgment, motion prediction, fine search and template protection; constructing a target distance mapping function fusing geometric distortion correction and attitude compensation, and realizing high-precision distance estimation based on monocular vision; laser radar point cloud information is integrated, an obstacle threat degree model is constructed, and cooperative path planning of following and obstacle avoidance is realized in combination with an improved TEB algorithm; and designing a linear velocity control law and an angular velocity control law based on the distance deviation and the azimuth angle deviation, and driving the robot to complete target following motion. According to the invention, high-precision and robust following of the robot to the target can be realized.

The invention belongs to the technical field of visual angle follow-up vehicle panoramic look-around, and particularly relates to a visual angle follow-up vehicle panoramic look-around system based on dynamic prediction, which is characterized in that monocular vision and laser radar time sequence data are fused, and a dense and stable dynamic target depth sequence is generated through dynamic target initial perception, cross-modal time sequence alignment and inter-frame depth correlation modeling; through environment interaction modeling and target behavior intention prediction double-branch parallel processing, a dynamic self-adaptive accurate track is output in combination with a weighted fusion strategy, and the prediction weight dynamic adjustment requirement of a complex motion scene is adapted; after a virtual view angle image is preliminarily synthesized based on the predicted trajectory, deviation detection and coordinate dynamic correction are realized through inter-frame target consistency verification, and artifact area positioning and pixel-level restoration are synchronously completed; multi-modal preprocessing and algorithm tasks are deployed according to hardware, and a model lightweight optimization and inter-frame prediction result multiplexing strategy is combined, so that low-delay operation of the system in a high-resolution panoramic image output scene is ensured.

The invention discloses a rapid robust monocular vision inertial positioning method and system. The method comprises the following steps: configuring IMU and monocular camera sensor parameters; preprocessing the IMU data; iMU attitude initialization is carried out; image features of the monocular camera are extracted, abnormal matching points are removed, and IMU pre-integration is carried out; monocular vision initialization is carried out; odometer attitude optimization: judging whether the feature points are mismatched or not based on an IMU pre-integration result, constructing a feature point re-projection error jacobian matrix, and performing blocking and diagonalization processing on the re-projection error jacobian matrix to respectively optimize inverse depths and image frame attitudes of the feature points; selecting a key frame based on a key frame identification rule after the current frame attitude update is completed; and judging according to the current frame, and removing a certain frame in the sliding window to reserve a sliding window space for adding the latest frame. According to the method, the inverse depth of the feature points and the image frame attitude are optimized respectively based on the counterweight projection error Jacobian matrix partitioning and diagonalization processing, and the calculation complexity is reduced.

The invention discloses a PCB soldering paste printing three-dimensional defect detection system based on multispectral imaging, and relates to the field of machine vision detection, and the system comprises a multispectral image acquisition module, a four-channel industrial camera based on RGB and near infrared, and a tunable LED light source array; acquiring reflection characteristic data under different penetration depths through a wavelength switching mechanism; the motion control trigger module is used for realizing positioning based on an XYZ three-axis objective table driven by a servo motor in cooperation with feedback of an encoder; a pulse width modulation signal is adopted to coordinate the moving speed of a camera shutter and a platform; according to the method, RGB and NIR four-channel imaging is combined with Beer-Lambert law modeling, so that double verification of material component quantitative analysis and three-dimensional shape reconstruction is realized, and limitation of monocular vision is avoided; structured light projection and binocular stereo matching technologies are adopted, cross-frame data alignment is realized in cooperation with an ICP algorithm, and detail defects can be better detected.

The invention provides a 3D anti-collision detection method and system based on monocular vision, and the method comprises the steps: obtaining the image data of a front environment collected by a vehicle-mounted monocular camera in real time, and carrying out the preprocessing of the image data, and obtaining the standardized image data; generating an optimized absolute depth map by fusing a first depth estimation result and a second depth estimation result based on the standardized image data; detecting a 3D obstacle based on the optimized absolute depth map and scene semantic information to obtain related information of at least one target obstacle; and performing collision risk assessment according to the related information, and triggering early warning when the risk meets an early warning condition. The invention provides a high-precision and high-robustness monocular 3D anti-collision detection scheme, monocular depth estimation based on deep learning, semantic segmentation and a dynamic safety model are combined, and an end-to-end anti-collision system with physical significance is formed.

The present application relates to a kind of unmanned aerial vehicle impact interception methods based on monocular vision, the azimuth measurement information and angle measurement information of target unmanned aerial vehicle are calculated by target detection algorithm based on monocular vision;Azimuth measurement information and angle measurement information are used, through orthogonal projection matrix and algebraic transformation, non-linear measurement is converted into pseudo-linear form, then pseudo-linear adaptive Kalman filtering algorithm is carried out to estimate the position of target unmanned aerial vehicle, finally, unmanned aerial vehicle maneuvering strategy is used to control impact unmanned aerial vehicle to realize the impact interception of target unmanned aerial vehicle.The method of the present application is obviously improved in system noise robustness, target state estimation accuracy and pursuit efficiency.The present application also relates to impact unmanned aerial vehicle guided by the above impact interception method.

The invention discloses a monocular vision inertial fusion positioning method based on an optical flow smoothness constraint, which comprises the following steps: designing a visual feature enhancement algorithm based on the optical flow smoothness constraint, improving feature point tracking and elimination in a visual inertial navigation system, and carrying out tight coupling optimization on a visual residual error part so as to improve the quality of observed quantity in positioning; comprising the following steps: generating a clustering result and a support factor for each feature point tracking vector of an acquired image through optical flow smoothness constraint; constructing a compound expectation maximization (EM) algorithm, fusing a support factor into algorithm solution to serve as prior information, and eliminating wrong matching point pairs to obtain a correct matching vector; adaptively adjusting the visual weight of the pose estimator, and performing region shielding; and monocular vision inertial fusion positioning based on optical flow smoothness constraint is realized.

The invention relates to the technical field of machine vision, and provides a concrete vibration operation real-time guidance method, device and equipment and a storage medium, and the method comprises the steps: generating a digital map and a monitoring grid with a physical proportion relation, dynamically detecting the change of a concrete bin surface in construction, and automatically updating the digital map and the monitoring grid, thereby achieving the real-time guidance of the concrete vibration operation. Vibrating personnel and a vibrating rod are recognized through an instance segmentation model to judge whether the vibrating rod is in an effective vibrating state or not, equivalent vibrating points are extracted and positioned to a digital map, the vibrating coverage amount in a monitoring grid is counted and compared with a preset construction standard, and an under-vibration or over-vibration area is automatically recognized; and further generating directional guide information or early warning prompt information with spatial directivity. According to the method, self-adaptive perception and accurate operation guidance of the dynamic construction surface are realized by a low-cost monocular vision scheme.

The invention discloses a digital eyestrain detection method based on multi-task learning and an intelligent terminal. The method mainly solves the problems that the prior art depends on a single mode, only pays attention to a single task (for example, only the opening and closing state is detected), and comprehensive evaluation on the fatigue state of a user, especially the eye dryness condition is lacked. The invention provides a digital eye fatigue detection method based on multi-task learning and an intelligent terminal, and aims to realize high-precision fatigue and distance detection on low-cost hardware through fusion of multi-task learning and a monocular vision geometric model, solve the problems of limited precision of a single mode and redundancy of separate deployment calculation in the prior art, and improve the detection precision of the eye fatigue. The method is suitable for mobile terminals such as smart phones and tablet computers. After the scheme is deployed on the intelligent terminal, a user can be helped to keep a healthy screen use distance and a healthy screen display parameter, a healthy blinking habit is maintained, and dry eyes, astringent eyes and eyestrain are reduced.

The application discloses a 4D video generation method fusing monocular vision and terrain elevation data, comprising the following steps: acquiring video frame images and corresponding camera pose data collected by a monocular camera; acquiring terrain elevation data of a target area, and constructing a three-dimensional elevation network model; based on the camera pose data, establishing a ray projection model for pixels in the video frame images, calculating the intersection of the rays and the three-dimensional grid model, and obtaining the reference distance from the monocular camera to the ground surface corresponding to the pixels; generating an initial relative depth map of the video frame images by using a depth estimation model; selecting high-confidence ground pixels from the video frame images as reference points, taking the reference distance corresponding to the reference points as the true value, and performing absolute scale correction on the initial relative depth map to obtain an absolute depth map; performing back projection calculation on the absolute depth map, the camera pose data and pixel color information to generate a three-dimensional point cloud, mapping the three-dimensional point cloud to a real geographic coordinate system, and generating 4D video data in time sequence.

This application discloses a method for constructing a dynamic risk map of a vehicle driving environment based on monocular vision, relating to the field of image recognition technology. It addresses the low accuracy and reliability of the final risk map output in existing monocular vision perception technologies due to phenomena such as target jitter, depth hovering, and repeated counting. The method analyzes the ground contact state, contact point coordinates, and foot depth of each target based on a reference support domain. It then projects, temporally fuses, and spatially deduplicates the contact point coordinates based on a projection strategy and foot depth to obtain a set of obstacle candidate points. Furthermore, it constructs a BEV occupancy grid map and a BEV risk grid map. Finally, it generates an image domain risk map based on pixel risk data and the BEV occupancy grid map, and constructs a cognitive risk map. The method outputs the cognitive risk map, the image domain risk map, and the BEV risk grid map, thereby improving the accuracy and reliability of the final risk map output.

The present application relates to monocular vision detection field, specifically to a kind of monocular vision 3D target detection method, low-level and high-level semantic features of picture are extracted using network skeleton and feature pyramid structure, the approximate position of target is obtained using candidate network P-Net, then the center point position of target is accurately obtained by refining network O-NeT, in order to eliminate the position deviation introduced when the image is down-sampled, in the present application, the position deviation information of target center will also be output, in addition, in order to solve the problem that the distance depth information of target is not estimated accurately in monocular 3D target detection, a simple and effective joint context module is proposed in the present application to more accurately predict the depth distance information of target, the problem of balancing comprehensive accuracy, detection speed and landing cost in the present monocular vision 3D target detection is solved.

The invention provides an unmanned aerial vehicle control system based on monocular vision. The unmanned aerial vehicle control system comprises a monocular acquisition device for acquiring a target image; the laser range finder detects a first distance from the unmanned aerial vehicle to the ground; the flight control module detects a second distance from the unmanned aerial vehicle to the ground and inertial data, and controls launching of the steering engine launching device according to the control signal or controls tracking of the unmanned aerial vehicle according to the tracking position; the visual perception module performs target detection on the target image by using a deep neural network; extracting multi-scale fusion features of the image features by using a decoder; utilizing a first perceptron to extract inertial characteristics of the inertial data; performing fusion and scale transformation on the multi-scale fusion feature and the inertial feature to obtain a third distance; the control module determines the target distance between the unmanned aerial vehicle and the target object according to the first distance, the second distance and the third distance, determines the target position of the target object according to the target distance, the target detection result and the coordinate mapping relation, and outputs a control signal or a tracking position according to the target position.

The invention discloses an automatic calibration and data processing method and tool for a monocular vision camera, and the method comprises the steps: controlling the camera to collect an initial calibration image in response to a one-key calibration instruction, and carrying out the quality verification to obtain a target calibration image; based on the ground feature points of the target calibration image, synchronously solving equivalent variable focal length model parameters and a downward inclination angle value through an integrated calculation process; establishing and solving a spatial distribution compensation model by using the equivalent variable focal length model parameter, the downward inclination angle value and the world coordinate value of the ground feature point to obtain an error compensation coefficient; packaging the equivalent variable focal length model parameter, the downward inclination angle value and the error compensation coefficient into a structured configuration file, and uploading the structured configuration file to a cloud server; and receiving an update instruction of the cloud server, and performing hot update on the error compensation coefficient. According to the method, automatic calibration of zero hardware addition, one-key completion and cloud synchronization can be realized, and the efficiency and the intelligent degree of monocular vision camera calibration are improved.

The invention belongs to the technical field of computer vision and automatic control, and particularly relates to a two-degree-of-freedom optical tracking method and system based on monocular vision, a calibration plate is arranged on a scanner, and a single calibration image containing the calibration plate is shot through a monocular camera; calibration point coordinates of the calibration image are extracted, the calibration image is input based on an ideal camera imaging model, the physical spacing of each calibration point, the image resolution, the pixel size and the lens focal length are known parameters, and the principal point coordinates, the pixel size ratio, the principal distance and the z-direction translation amount of the camera are estimated; introducing a comprehensive lens distortion model containing radial distortion, eccentric distortion and thin prism distortion, and obtaining an accurate radial distortion coefficient, an eccentric distortion coefficient, a thin prism distortion coefficient and rotation and translation amounts of the camera through decoupling calculation; theoretical pixel coordinates of the calibration points are obtained through reverse re-projection, and calibration precision is evaluated and optimized; the camera visual angle is adjusted through the holder control system, so that the scanner is located in the center area of the camera visual field.

The invention discloses a monocular microscopic imaging surface micro-nano defect 3D reconstruction detection device and method.The detection device comprises an illumination imaging module, a displacement module and a laser ruler distance measuring system; the displacement module comprises a shaft guide rail, a shaft guide rail, a self-rotating shaft, a shaft guide rail, a shaft guide rail, an objective table and a shaft guide rail arranged above the objective table, the shaft guide rail is mounted on the swing shaft; the illumination imaging module is mounted on the shaft guide rail and comprises an electric zoom imaging module and stroboscopic light sources which are annularly and uniformly arranged around the electric zoom imaging module; the illumination angle and the illumination aperture angle of each stroboscopic light source are electrically controlled and adjustable; the laser ruler distance measuring system is used for measuring the real moving distance of the shaft guide rail. According to the invention, on the premise that the detection efficiency and precision are ensured, micro-nano defect detection can be realized only by monocular vision, and 3D reconstruction is realized.

This invention discloses a video tracking and recognition method for the spatial trajectory of hoisted objects based on multi-source data fusion, belonging to the field of machine vision monitoring technology. The method includes acquiring pre-loaded physical boundary constants of the tower crane and simultaneously collecting multi-source data during tower crane operation. Based on the physical boundary constants, kinematic calculations are performed on the multi-source data to obtain the relative vertical displacement, the three-dimensional absolute spatial coordinates of the luffing trolley center point, and the absolute coordinates and three-dimensional rotation matrix of the camera's optical center in the global three-dimensional coordinate system. Edge detection is performed on the temporally aligned color video frame matrix to extract candidate wire rope contours and generate a binary edge pixel matrix. Collinear edges are mapped to polar coordinate space using Hough transform. This invention reduces the uncertainty of monocular vision depth scale and improves stability and accuracy under conditions of large-amplitude lifting and complex swinging.

This invention discloses a method for generating foggy images based on monocular vision depth estimation. Aimed at applications in foggy image simulation, this method employs an improved physical model and an improved depth map estimation model to process fog-free images and simulate realistic foggy images, meeting the application needs of image processing research such as image dehazing and target detection in foggy images. First, this invention uses a monocular vision depth estimation network to generate an initial relative depth map of the scene. Then, considering the characteristics of foggy images, it uses image super-resolution and edge smoothing techniques to optimize the initial depth estimation map, obtaining an optimized depth map of the scene. Finally, based on the optimized depth map, an improved transmittance generation model is used to generate a transmittance map of the scene. Finally, the improved transmittance map is fused with the scene image to obtain a realistic foggy image. This invention can be applied to fields such as autonomous driving simulation, forest fire prevention, remote sensing positioning, and virtual reality.

The invention discloses an unmanned aerial vehicle monocular vision inertial positioning method based on under-forest geometric representation, and belongs to the technical field of forestry monitoring. In order to solve the problem of positioning drift of the unmanned aerial vehicle caused by background pseudo observation in a complex under-forest environment, a lightweight instance segmentation network is adopted to carry out trunk instance segmentation and mask generation on an input image, and a hard semantic mask of a trunk area is obtained; constructing a visual residual weighting function which comprises a full-region point feature constraint and a line feature constraint, and performing optimization processing by using a soft weighting function to obtain an input image after semantic mask cutting optimization; geometric constraints of the trunk line segments are set, and candidate trunk line segments after geometric constraint processing are obtained; in a sliding window nonlinear optimization framework, uniformly encoding point feature constraints and line feature constraints into an information matrix; and optimizing all to-be-optimized variable sets in the sliding window by using a weighted least square method, then adjusting the scale of the sliding window through a marginalization strategy, and outputting an optimized global trajectory.

The application discloses a monocular vision-based material motion behavior monitoring method and system, relates to the field of material motion monitoring, and first utilizes a monocular high-speed camera to collect a material motion video, adopts instance segmentation and time sequence tracking technology to extract a continuous state sequence of the material on a two-dimensional image plane. The core lies in introducing a time sequence deep learning model to analyze the state sequence, automatically identifying a key event of collision between the material and a working surface of equipment, so that a collision moment and a contact point coordinate are accurately acquired. Thus, the collision contact point is used as a geometric anchor point for three-dimensional space inverse calculation, the two-dimensional image trajectory is reversely projected and reconstructed into a three-dimensional space motion trajectory in combination with camera internal parameters and physical parameters of a device reference surface, and then key dynamic parameters such as speed and a restitution coefficient are solved, so that low-cost, high-precision automatic online monitoring is realized.

This invention belongs to the field of vision and image processing technology, and provides a method, system, device, and medium for detecting dangerous actions based on radio frequency (RF) images. The method includes: acquiring a monocular video stream of a target scene, reconstructing a three-dimensional digital space geometric model of the target scene, and solving the pose matrix; collecting raw multidimensional feature data emitted by a signal transmitting device, mapping the raw multidimensional feature data to the three-dimensional digital space geometric model, and generating a three-dimensional scalar field data volume characterizing human radio frequency behavior disturbances; calculating a dynamic disturbance residual field based on the three-dimensional scalar field data volume, projecting and encoding the dynamic disturbance residual field to generate a 2.5D multi-channel human radio frequency behavior feature image, and identifying dangerous actions. This invention utilizes monocular vision to reconstruct a three-dimensional digital space and solve the pose of RF devices, establishing a physical space-digital space mapping relationship, and eliminating the strong dependence of traditional RF solutions on specific room layouts and multipath effects.

The application discloses a kind of fusion ToF and monocular vision's robot joint layer perception system, belong to robot perception field, system uses Eye-in-Hand architecture, integrates ToF camera and RGB camera;Through joint bilateral filtering and Kalman filtering, improve depth map quality;Based on SURF feature matching and the ICP registration of KD-Tree acceleration is realized target six degree of freedom pose estimation;Combined with improved RRT algorithm and hierarchical collision detection planning capture path.The robot joint layer perception system of fusion ToF and monocular vision provided in the application, through the space alignment and fusion of two kinds of visual information, make up for respective limitations, improve the robustness and precision of overall perception system;In single-arm and double-arm capture experiment, system positioning accuracy reaches ±3mm, dynamic scene capture success rate is high.It is suitable for industrial automation, logistics sorting and the like.

This invention relates to the field of nuclear power plant equipment replacement and discloses a monocular vision alignment method applicable to reactor detector assemblies. First, the image coordinates of the reactor detector assembly's center are acquired. Then, based on the distance between the plane where the camera is located and the plane where the reactor detector is located, the world coordinates of the reactor detector assembly's center are obtained. Next, based on the distance between the world coordinates of the reactor detector assembly's center and the camera's optical axis, the camera movement is adjusted. Once the camera's optical axis coincides with the center of the reactor detector assembly, the center of the mechanical gripper is adjusted to coincide with the center of the reactor detector assembly based on the coordinate offset between the camera's optical axis and the center of the mechanical gripper. This invention eliminates the need for external parameter calibration by placing calibration objects inside the reactor, and the method is simple, highly feasible, and very suitable for reactor detector assembly replacement scenarios in nuclear power plants.

The invention provides a monocular vision vehicle three-dimensional size estimation method and system based on multi-stage feature fusion, and relates to the field of computer vision. Comprising the following steps: detecting a plurality of vehicles in an image collected by a monocular vision system by adopting a target detection algorithm to obtain a plurality of two-dimensional bounding boxes; a main vehicle is screened and determined, and pixel coordinates and deep semantic feature vectors of vehicle body contour key points are obtained in the two-dimensional bounding box by adopting a key point detection model; according to internal and external parameters of a monocular vision system and a back projection theory, three-dimensional world coordinates corresponding to key points of a vehicle body contour are calculated, and geometric estimation of the three-dimensional size of a host vehicle is carried out to obtain a prior value; and splicing the feature vector formed by the prior values, the feature vector of the vehicle body contour key point and the deep semantic feature vector to obtain a comprehensive feature vector, inputting the comprehensive feature vector into a network model for size regression, and obtaining an estimated value of the three-dimensional size of the host vehicle. According to the invention, the three-dimensional size of the vehicle can be calculated with high precision and low cost based on monocular vision.

The invention relates to the technical field of vehicle charging, in particular to an automobile charging robot charging port positioning method and device based on monocular vision, and the method comprises the following steps: S1, continuously collecting multiple frames of RGB images for the same charging port in a static state of a mechanical arm through a monocular camera connected to the tail end of the mechanical arm; s2, inputting into a deep learning model obtained by pre-training, and synchronously outputting a soft segmentation mask of each jack area of the charging port, a circle center pixel coordinate corresponding to each jack and a predicted pose vector from a camera coordinate system to a charging port coordinate system; s3, weighted fusion is carried out on the predicted pose vectors output by the multiple frames of RGB images respectively, and optimized pose vectors are obtained; and S4, on the basis of the optimized pose vector and a pre-calibrated hand-eye relation, the target motion pose of an end effector of the mechanical arm relative to the charging port is solved, and the mechanical arm is controlled to execute the gun insertion action. The cost effectiveness and the real-time performance can be considered.

This invention discloses a 3D train detection and positioning method based on ground-based monocular vision. It eliminates the need for satellite signals or additional trackside equipment. Through convolutional-trans depth estimation and pseudo-point cloud optimization, the absolute relative error of depth estimation can be reduced to below 0.06. After pseudo-point cloud optimization, the 3D detection accuracy is improved by more than 20%, achieving a final positioning accuracy of 1 meter, meeting the auxiliary positioning requirements of train control systems. Furthermore, through view frustum region extraction and lightweight network design, the inference frame rate is ≥10fps, meeting the real-time positioning requirements of high-speed trains. In addition, existing ground-based monocular cameras can be reused, eliminating the need for additional hardware. The deployment and maintenance costs are only 1 / 10 of those of transponder positioning systems. Overall, this invention provides a novel technical path for train positioning and can be widely applied to auxiliary positioning systems for conventional and high-speed railways. It is particularly suitable for the intelligent upgrading and transformation of existing lines, possessing extremely high engineering application value.

The application discloses a PCB solder paste printing three-dimensional defect detection system based on multispectral imaging, and relates to the field of machine vision detection.The technical scheme points of the application include a multispectral image acquisition module, a four-channel industrial camera based on RGB and near-infrared, and a tunable LED light source array; the reflection characteristic data under different penetration depths are acquired through a wavelength switching mechanism; a motion control trigger module is based on a servo motor driven XYZ three-axis object table, and positioning is realized in cooperation with an encoder feedback; a pulse width modulation signal is adopted to coordinate the camera shutter and the platform moving speed; the application realizes the dual verification of material composition quantitative analysis and three-dimensional topography reconstruction through the combination of the four-channel imaging of RGB and NIR and the Beer-Lambert law modeling, and avoids the limitations of monocular vision; the cross-frame data alignment is realized through the projection of a structured light and a binocular stereo matching technology in cooperation with an ICP algorithm, and the details defects can be better detected.

The invention belongs to the technical field of measurement, and relates to an unmanned aerial vehicle obstacle depth measurement method based on monocular vision and point transformation. According to the method, continuous images are acquired through a monocular camera carried on an unmanned aerial vehicle, and attitude angle data are acquired in combination with an airborne inertial measurement unit. Firstly, an obstacle target in an image is identified by using a trained target detection neural network; thirdly, extracting feature points in the target area by adopting a feature descriptor, and carrying out feature matching between adjacent frames to construct matching point pairs; and then, constructing a mapping mathematical model between the attitude angle and the image millimeter coordinate transformation, and performing coordinate correction on the matching point pairs based on the model. Generating and screening matched feature line segments based on the corrected feature point pairs, and calculating a corrected pixel length; and finally, a depth measurement model under a standard attitude is constructed, and accurate calculation of the depth of the obstacle is realized by using the correction pixel length. Experimental results show that accurate measurement of the target depth can be effectively realized under the conditions of standard attitude and attitude offset, and the method has high measurement precision and robustness.