766 results about "Fish eye" patented technology

The invention relates to a wharf sensing method and system based on data fusion of a laser radar and a camera, and the method comprises the steps: firstly employing a fisheye camera to shoot a checkerboard calibration plate at multiple angles, employing a Zhang's calibration method to calculate an internal reference matrix and a distortion coefficient of the fisheye camera, and synchronously collecting the data of the laser radar and the camera in real time; the method comprises the following steps: performing distortion removal processing on a camera image, calculating an external parameter matrix between a laser radar and a camera through a feature extraction algorithm, a feature point matching algorithm and a nonlinear optimization algorithm, and performing data fusion through an internal parameter matrix, the external parameter matrix and a bilinear interpolation method to generate fusion point cloud data with color information; and finally, identifying a key target of the wharf by marking and training a CNN model, and calculating three-dimensional parameters of the key target by adopting a specific calculation method, so that a sailor can quickly distinguish the target in the fused point cloud data, the accuracy and efficiency of wharf environment perception are remarkably improved, and accurate environment perception and decision support are provided for ship berthing and departing operation.

The invention discloses a data processing method of an intelligent cold stamping production line integrated with an MES system, and the method comprises the steps: S1, installing at least one fisheye lens for detecting a stamping workpiece, collecting data, preprocessing the data, and building a three-dimensional scene, S2, carrying out the further processing of the preprocessed data in S1 through a flexible sensor network, the method comprises the following steps: S1, carrying out cross-modal fusion and feature extraction on data, S3, transmitting the data extracted in S2 to a digital twinborn body for dynamic updating and prediction, S4, carrying out deformation with the assistance of quantum entanglement and carrying out over-distance early warning, and S5, carrying out data storage and knowledge management. According to the data processing method of the intelligent cold stamping production line integrated with the MES system, a digital twinborn dynamic updating and predicting technology is utilized, quantum-classical hybrid simulation modeling and reinforcement learning algorithms are combined, precise prediction and optimization adjustment can be conducted on mold abrasion, process parameters and the like, the production efficiency and the product quality are improved, and the production cost is reduced. The production cost and the energy consumption are reduced.

The invention relates to the technical field of automatic driving, and discloses a parking assisting method and system for automatic driving. The method comprises the following steps: cascading and splicing a first fisheye image acquired by a main camera in a vehicle body and a second fisheye image acquired by a slave camera to obtain a panoramic image; performing parking space boundary identification and obstacle analysis on the panoramic image to obtain parking space boundary information and obstacle position information, and synchronously performing distance calculation on a radar echo signal of an ultrasonic sensor to obtain distance measurement data; and a parking environment map is created according to the parking space boundary information, the obstacle position information and the distance measurement data, path planning is carried out based on the parking environment map, a parking execution instruction is obtained, and the parking execution instruction is used for vehicle parking auxiliary control. According to the method, the view coverage range and the flexibility of the observation angle are expanded, and the execution precision and safety of parking auxiliary control are improved.

The invention discloses a calibration method, system and device for a vehicle around-view camera and a computer medium, and relates to the technical field of image processing, and the method comprises the steps: collecting a calibration plate image obtained after a calibration plate is shot by the around-view camera on a vehicle; projecting the calibration plate image to a world coordinate system to obtain a distorted image; selecting an initial angular point in the distorted image according to the calibration plate; projecting the initial angular point to a fisheye image coordinate system of a look-around camera to obtain a target angular point; according to the target angular point, generating an initial external parameter of the look-around camera; in the calibration plate image, extracting overlapped corner point parameters in the overlapped shooting area; and based on the initial external parameter, generating a target external parameter which enables the error between the overlapping angular point parameters to be minimum, and taking the target external parameter as the external parameter of the surround-view camera. According to the method, the extraction precision and robustness of the angular points are improved, the continuity and smoothness of the images shot by the surround view camera are improved, splicing gaps and dislocation phenomena are reduced, and the calibration result of the vehicle surround view camera is more accurate.

The invention provides an image correction method and system based on a nonlinear distortion distribution elliptical fisheye lens, and a medium. The method comprises the following steps: establishing a nonlinear mixed distortion imaging model based on a mapping relation between image coordinates and a spherical direction; defining a distortion vector, and performing fitting processing on the distortion vector based on a two-dimensional B-spline function to obtain a distortion function; the method comprises the following steps: acquiring images in a five-surface three-dimensional angle calibration space based on an elliptical fisheye lens, analyzing lens imaging parameters, and generating parameter values of a distortion function; analyzing parameter values of the distortion function based on a nonlinear mixed distortion imaging model to generate calibration parameters, and remapping pixel points in the distorted image based on the calibration parameters; performing image reconstruction on the remapped distorted image to obtain a corrected image; improving the sampling density of a target area by constructing a nonlinear mixed distortion imaging model; the system adapts to complex monitoring environments such as galleries and entrances and exits; and spatial distortion is flexibly expressed through a B spline, so that the modeling capability is improved.

The invention provides a GNSS data quality prediction method and device based on a fisheye camera, and a medium, and belongs to the technical field of data processing, and the method comprises the steps: determining the satellite orbit information of a future time period with a to-be-evaluated observation station as a reference point based on ephemeris prediction; according to the fisheye sky image shot at the to-be-evaluated observation station and the satellite orbit information, determining satellite feature data of a future time period and the occlusion porosity of the future time period; respectively inputting the occlusion porosity of the future time period and at least part of satellite feature data of the future time period into a general prediction model and an observation station prediction model corresponding to the observation station to be evaluated for data quality evaluation, and obtaining a first data quality evaluation index and a second data quality evaluation index; and determining a data quality prediction result based on the first data quality evaluation index and the second data quality evaluation index. A general prediction model and an observation station exclusive prediction model are introduced to carry out double-model evaluation, and accurate prediction of GNSS observation data quality is realized.

The invention relates to the technical field of GNSS observation environment evaluation, in particular to a GNSS observation environment complexity measurement method, device and equipment and a medium, and the method comprises the steps: obtaining a sky image of a target area through a fisheye camera, and carrying out the recognition of the sky image through an image processing algorithm, and obtaining a sky recognition result; calculating trajectories of all satellites based on almanac information, and obtaining an openness index of the target region in combination with a sky recognition result; gNSS observation data are obtained, and a quality index is calculated according to the GNSS observation data; and combining the openness index with the quality index through a gradient lifting regression algorithm, predicting the calculation precision, and converting the predicted calculation precision into environment complexity. According to the GNSS observation environment evaluation method, the machine learning algorithm is combined with the sky view and the data quality index, the evaluation accuracy of the GNSS observation environment under the complex environment is improved, and the GNSS observation environment evaluation method is suitable for GNSS observation environment evaluation under various complex terrain conditions.

The invention provides a robot-based blade interior autonomous crawling sensing method and system, and the method comprises the steps: carrying a multi-sensor array on a robot, the multi-sensor array comprises a binocular fisheye camera array and a ToF sensor array, controlling the robot to start from a path travel starting point, obtaining a panoramic depth image based on the multi-sensor array, and transmitting the panoramic depth image to the robot; recognizing an obstacle based on the panoramic depth image, constructing a three-dimensional point cloud map based on the obstacle, avoiding the obstacle based on the three-dimensional point cloud map, generating a feasible path in the blade, and enabling the robot to move based on the feasible path until the robot reaches a path travel terminal point or a minimum space in which the robot cannot pass. Autonomous crawling perception of the robot in the blade can be ensured, movement of the robot in the blade is ensured, and detection is realized.

The invention discloses a coal mining machine positioning and visual analysis system under a multi-algorithm optimized fisheye camera, and relates to the field of coal mining machine positioning and visual analysis. The deployment acquisition module is installed according to an optimization algorithm, has an intelligent function and acquires images in a multispectral manner; the preprocessing module corrects distortion by using an improved model and enhances an image in combination with multiple algorithms; the identification calculation module integrates multiple technologies to identify features of the coal mining machine and accurately position the coal mining machine; the tracking and positioning module fuses multi-algorithm tracking, and deep learning is used for predicting a trajectory during shielding; the visual analysis module analyzes the state of the coal mining machine through multiple technologies; and the transmission interaction module transmits data by using a fusion technology and provides multiple interaction modes. The method is accurate in positioning, the fisheye camera and the correction technology improve the positioning accuracy, and accurate tracking can be achieved when shielding is carried out; visual analysis is deep, and the state of the coal mining machine and environmental hidden dangers are monitored; data transmission is safe and efficient, and interaction is convenient; the system has strong anti-interference and self-optimization capabilities, can adapt to an underground environment, and improves the coal mining operation efficiency and safety.

A basketball backboard interaction tracking system that captures, processes, and visually communicates data related to basketball interactions for enhanced player engagement and training. The system includes two configurations: an initial configuration with a transparent glass backboard positioned in front of a display and a second, modular configuration without an integrated display. The initial configuration uses various sensor types, including fisheye cameras, light curtains, microphones, and infrared or ultrasonic sensors, to track basketball interactions, with real-time feedback displayed on the integrated screen or a mobile interface. The second configuration, optimized for environments without a display, includes sensor modules with pressure sensors, RFID readers, or sound-triangulating microphones. Both configurations incorporate a controller unit for processing sensor data, a cloud or on-premises server for data storage, at least one power supply, and a user interface for real-time feedback. This system supports interactive training, game statistics, and data analysis for skill improvement.

The invention discloses an automatic parking obstacle detection method and device. A semantic segmentation model of each scene is trained by collecting multi-path look-around fisheye camera image sets of different scenes, so that a semantic segmentation model library is generated, and when parking is performed, a target semantic segmentation model is matched in the semantic segmentation model library based on a driving scene, so that the adaptability and accuracy of the semantic segmentation model in different scenes are improved, and the parking experience is improved. And the four fisheye images are subjected to semantic segmentation based on the four fisheye images and then spliced into a bird's-eye view angle, so that the problem that the bird's-eye view angle easily causes loss or distortion of obstacle information is avoided, an accurate travelable region boundary is provided, and the obstacle detection accuracy is improved.

The invention relates to a wafer pose recognition system and an image acquisition device and an image acquisition method thereof, the wafer pose recognition system comprises a support structure for supporting a wafer to be recognized and the image acquisition device arranged above the wafer to be recognized, and an image sensing module is arranged in the image acquisition device. A fisheye lens group or a negative lens is adopted to enlarge the visual field width of an image sensor, so that the image sensing module can obtain complete edge image information of a wafer in one-time imaging, the positioning efficiency is improved, an annular first light shielding plate and a circular second light shielding plate are arranged on an optical path of the lens to form an annular visual field in a matched mode, and the positioning precision is improved. The edge blurred part image and the middle too bright part image are physically excluded, the imaging quality in the annular view is improved, the clear and complete wafer positioning edge image can be obtained, the contour recognition precision during image recognition is improved, and the positioning accuracy is improved.

This application provides a fisheye image processing method, an electronic device, and a computer-readable storage medium. The method includes: acquiring a fisheye image; generating a grayscale image according to the fisheye image; generating a first gradient image in a horizontal direction and a second gradient image in a perpendicular direction; generating an azimuthal image according to the first gradient image and the second gradient image; and determining an effective imaging area according to the azimuthal image. The anti-distortion correction for the fisheye camera is performed in the process of shooting images according to the effective imaging area of the fisheye camera, and thus the imaging area of the fisheye camera is accurately calibrated when the images are shot by the fisheye camera.

The utility model provides a high-pixel confocal panoramic fisheye optical system and a camera module applying the same, the high-pixel confocal panoramic fisheye optical system is mainly composed of seven lenses, the first lens has negative focal power, the object side surface of the first lens is a convex surface, and the image side surface of the first lens is a concave surface; the second lens has negative focal power, and the image side surface of the second lens is a concave surface; the third lens has negative focal power, the object side surface of the third lens is a concave surface, and the image side surface of the third lens is a convex surface; the fourth lens has positive focal power, and the object side surface of the fourth lens is a convex surface; the fifth lens has positive focal power, the object side surface of the fifth lens is a convex surface, and the image side surface of the fifth lens is a convex surface; the sixth lens has negative focal power, the object side surface of the sixth lens is a concave surface, and the image side surface of the sixth lens is a concave surface or a convex surface; the seventh lens has positive focal power, the object side surface of the seventh lens is a convex surface, and the image side surface of the seventh lens is a concave surface or a convex surface; through reasonable matching of the lens shape and the focal power, the lens has the advantages of high pixel, infrared confocal, ultra-wide angle and large aperture, the performance requirements of large aperture and high pixel are met, and the imaging effect of equipment matched with the system is further improved.

A method to fine-tune extrinsic parameters of a fisheye lens applied to surround view splicing is provided. Fisheye images are obtained from multiple fisheye lenses. Both sides of the fisheye images contain calibration boards. Initial extrinsic parameters are obtained based on the fisheye images. Multiple corner points of the calibration boards in the fisheye images are detected. The corner points are arranged so that the corner points are connected into a quadrilateral in sequence. The initial extrinsic parameters are multiplied by a fine-tuning value to obtain intermediate extrinsic parameters. The corner points are projected to a top-view projection by using the intermediate extrinsic parameters. The intermediate extrinsic parameters are adjusted according to the shape, size, and overlapping effect of the area surrounded by the corner points in the top-view projection.

The invention provides a three-dimensional target detection model training method, a three-dimensional target detection method and equipment, and the method comprises the steps: obtaining a sample image shot by a fisheye camera, and marking the three-dimensional attribute marking information of a to-be-detected target, comprising a first direction angle, a first division area to which the first direction angle belongs and a first deviation angle in the first division area. And inputting the sample image into the initial detection model, and obtaining three-dimensional attribute prediction information of the to-be-detected target, including a second direction angle, a second division area to which the second direction angle belongs and a second deviation angle in the second division area. And determining a loss function value based on the three-dimensional attribute prediction information and the three-dimensional attribute labeling information. And training the initial detection model based on the loss function value to obtain a three-dimensional target detection model. By marking the first division area to which the first direction angle belongs, the error of the predicted second direction angle is reduced, and the accuracy of the three-dimensional target detection model is improved.

The invention discloses a distortion correction method and device for a fisheye image, and relates to the technical field of computer vision. Obtaining a query vector of the to-be-corrected fisheye image from the query vector set according to the distortion degree of the to-be-corrected fisheye image; according to the query vector of the to-be-corrected fisheye image, determining control conditions of the to-be-corrected fisheye image in multiple levels; the control condition of each level corresponds to each modulation block in the level correction network in sequence; and inputting the original features of the to-be-corrected fisheye image and the control conditions of the to-be-corrected fisheye image in each level into the level correction network, fusing the original features and the corresponding control conditions by using the controllable convolution modulation block, and carrying out correction on the to-be-corrected fisheye image in the controllable attention modulation block. And global enhancement is carried out on the features output by the controllable convolution modulation block through a control condition, and a distortion correction image of the fisheye image to be corrected is obtained. The method is high in generalization ability for different distortion conditions, and the distortion correction precision of the fisheye image can be improved.

The invention relates to an intelligent ship berthing and leaving self-positioning method and system based on texture point clouds, and the method comprises the steps: firstly employing a fisheye camera to shoot a checkerboard calibration plate at multiple angles, employing a Zhang's calibration method to calculate an internal reference matrix and a distortion coefficient of the fisheye camera, and synchronously collecting the data of a laser radar and the camera in real time; the method comprises the following steps: carrying out orthodontic processing on a camera image, calculating an external parameter matrix between a laser radar and a camera through a feature extraction algorithm, a feature point matching algorithm and a nonlinear optimization algorithm, and carrying out data fusion through an internal parameter matrix, the external parameter matrix and a bilinear interpolation method to generate texture point cloud data with color information; the method comprises the following steps: performing texture point cloud registration through an improved NDT-color algorithm in cooperation with open-loop mileage estimation to preliminarily determine the position change of a ship, optimizing a graph model by using a closed-loop detection and graph optimization technology, generating an optimized global displacement vector, and further generating a wharf high-precision map containing color texture information. And real-time self-positioning of the ship in a wharf high-precision map is realized.

The invention relates to a scanning mapping method and device, and belongs to the technical field of mobile mapping and three-dimensional reconstruction, the method is applied to a scanning mapping device, the device comprises a laser radar, an IMU and a fisheye camera, and the method comprises the following steps: obtaining point cloud data obtained by scanning of the laser radar at each moment, measurement data of the IMU and image data shot by the fisheye camera; determining pose data of the scanning and mapping device at each moment according to the point cloud data, the measurement data and the image data at each moment; performing coordinate conversion processing on the point cloud data at each moment according to the pose data at each moment to obtain coordinates of the point cloud data in a global coordinate system; and constructing a target map according to the coordinates of the point cloud data in the global coordinate system. The pose of the scanning mapping device is determined according to the observation data of the laser radar, the IMU and the fisheye camera, and the pose robustness can be improved. And mapping is carried out according to the pose, so that the mapping accuracy can be improved.

The invention discloses a wafer nanoscale defect detection system and method based on multi-mode and feature fusion, the system comprises a multi-band LED light source group, a CMOS camera, a telecentric imaging lens and a displacement platform, the multi-band LED light source group is arranged at the uppermost part, and the CMOS camera, the telecentric imaging lens and the displacement platform are sequentially arranged below the multi-band LED light source group; the multiband LED light source group comprises an angle gt; an incident light LED with the angle of 70 degrees, an incident light LED with the angle of 10-30 degrees, a coherent light source and a phase shift module; the CMOS camera shoots a wafer picture, and the telecentric imaging lens captures details on the surface of the wafer; the displacement platform bears the wafer and carries out displacement. According to the method, the limitation of two-dimensional detection is broken through, and on the premise of not sacrificing the detection speed, high-sensitivity, high-precision and quantifiable online automatic detection of fisheye, pinhole and segment difference defects is realized through an innovative optical system design, a quasi-three-dimensional morphology reconstruction technology and an advanced deep learning network architecture.

The embodiment of the present invention discloses a method, device, equipment and storage medium for distortion correction of a vehicle-mounted fisheye camera. The method includes: obtaining first position information of key points in a calibration image; controlling a vehicle-mounted fisheye camera to collect the calibration image to obtain a distorted image, and extracting second position information of the key points in the distorted image; fitting the first position information and the second position information of the key points using a cubic B-spline curve fitting algorithm to obtain a correction relationship; obtaining the image to be corrected collected by the vehicle-mounted fisheye camera; and correcting the image to be corrected according to the correction relationship. The method for distortion correction of a vehicle-mounted fisheye camera provided by the embodiment of the present invention uses a cubic B-spline curve fitting algorithm to fit pixel points in a calibration image and pixel points in a distorted image to obtain a correction relationship, thereby performing distortion correction on the image collected by the fisheye camera according to the correction relationship, which can improve the accuracy of distortion correction.

The invention relates to the technical field of vehicle visual perception, and particularly discloses a multi-fisheye camera aerial view construction method and system for automatic driving, and the method comprises the steps: carrying out the offline calibration of each fisheye camera based on a polynomial fisheye imaging model, constructing a pixel back projection lookup table, and carrying out the offline calibration of each fisheye camera; performing table lookup type distortion removal processing on the acquired fisheye image; extracting multi-region checkerboard angular points, and establishing a cross-view-angle corresponding point set; solving an initial homography matrix from each fisheye camera to the aerial view reference plane, constructing a joint objective function including a checkerboard re-projection error, an overlapping region alignment error and a physical scale constraint, and performing global joint optimization on all the homography matrixes; generating each path of aerial view sub-graph by adopting a reverse mapping strategy; according to the method, the calibration process is visual, the geometric accuracy, the real-time performance and the engineering transportability are high, and the method is suitable for automatic parking, a vehicle looking-around system and automatic driving environment perception.

The invention discloses an anti-halo interference fisheye image effective area detection method, which performs modeling simulation on halo interference possibly existing in a fisheye image shooting process aiming at halo intensity, fuzzy degree, color, position and size factors, and constructs a fisheye image effective area detection model based on a deep neural network. The input of the model is a halo interference fisheye image, the output is the center coordinate and size of an effective area, and the anti-halo interference fisheye image effective area detection model comprises a fisheye image type identification module, an effective area identification module and an effective area parameter extraction module. The fisheye image effective area detection method provided by the invention has stronger robustness and higher detection precision for various halo interferences, can effectively solve the problems existing in the traditional method, and further improves the distortion parameter estimation precision and the distortion correction effect of the fisheye image.

The invention discloses an occupation grid map generation method fused with fisheye passable space detection, and the method comprises the steps: fusing the fisheye passable space detection with laser radar passable space detection, and generating a more accurate and stable passable space occupation grid map; the fisheye image information is rich, rain, fog, dust and the like are not detected to be impassable, the grids updated by laser detection are reversely projected into the fisheye image, the occupation probability of the grids detected to be impassable by the fisheye is reduced, laser false detection is inhibited, and the influence that the dust, the rain, the fog and the like are falsely detected to be impassable by the laser is reduced; meanwhile, the horizontal and vertical field angles of view of the fisheye camera are 180 degrees, so that no blind area exists basically, the passable space condition in the laser blind area can be detected, the occupancy probability of grids which are located in the blind area and are detected to be impassable by the fisheye is maintained, and the safety of automatic driving is guaranteed.

The embodiment of the invention provides a multi-view-angle-based stage fusion panoramic look-around image generation method and device, equipment and a computer readable storage medium. The method comprises the following steps: acquiring a target image through a target fisheye camera; constructing a multi-view spherical projection model according to the target image; and according to the spherical projection model, generating a target visual angle panoramic spliced image. In this way, the problems of geometric distortion, obvious edge seams, inconsistent brightness, texture information loss and the like existing in a traditional double-fisheye image splicing method are solved.

The invention relates to the technical field, and discloses a fisheye test needle group for testing a circuit board, which comprises a test board, the test board comprises a heating pressing block, an insulating component and a base, a test sample is placed between the heating pressing block and the insulating component, the test sample comprises a sample body and fisheye needles extending from pins on the sample body, and the fisheye needles are arranged on the base. A plurality of groups of double-end probes and elastic sheets are mounted in the base and are in butt joint with fisheye needles, the fisheye needles, the double-end probes and the elastic sheets form a test needle group, when the test sample sinks, the fisheye needles are in butt joint with the double-end probes, the elastic sheets are in butt joint with the fisheye needles to form two pairs of contacts, and when the test sample sinks to guide a first pair of contacts to appear, the first pair of contacts are in butt joint with the fisheye needles. And the wedge-shaped surface in the sinking direction guides the elastic sheet to incline so as to form the second pair of contacts, so that the current and voltage required by Kelvin testing can be met, the two pairs of contacts and a plurality of testing points are integrated in a single structure, and the use requirement is met by the appearance of lateral force while vertical downward force exists.

The present invention generally relates to a technology for recognizing objects in a Around View Monitoring (AVM) video for a vehicle. In particular, the present invention relates to a DNN-based object recognition method for multi-channel fisheye images of a vehicle AVM apparatus, by projecting multi-channel unit images of fisheye-lens cameras of the vehicle AVM apparatus onto a cylindrical projection plane so as to obtain multi-channel unit projection images, combining the multi-channel unit projection images on a single image so as to obtain a composite projection image, and inputting the composite projection image to a Deep Neural Network (DNN) model so as to obtain object recognition output. The invention has an advantage that AVM apparatus can effectively assist in safe driving by monitoring objects around the vehicle. Further, the present invention has an advantage that object recognition may be effectively achieved by the insufficient computing resources of vehicles.

The invention discloses an image retrieval method and device, electronic equipment, a storage medium and a product. The method comprises the following steps: constructing text description according to a text category; according to the text description, performing retrieval in a database by using a text encoder after fine tuning training to obtain a first target image; wherein the database comprises a plurality of fisheye images and image feature vectors of the fisheye images, the image feature vectors of the fisheye images are obtained through the image encoder after fine-tuning training, and the image encoder after fine-tuning training is obtained by performing fine-tuning training on the image encoder based on the fisheye images. According to the method, the image encoder is subjected to fine tuning training based on the fisheye image, so that the image encoder subjected to fine tuning training has the capability of accurately identifying the image feature vector in the fisheye image, the accuracy of text description and image matching can be improved, and the accuracy of image retrieval can be improved.

The invention relates to the technical field of image processing, and provides a fisheye image real-time processing method and device, a fisheye camera and a storage medium. When the fisheye camera is started, a coordinate conversion relation table from spherical coordinates to image coordinates is initialized and established according to a distortion table / distortion coefficient, a region of interest is determined according to a preset field angle, and a first spherical coordinate table is obtained according to the image coordinates of the region of interest and the distortion table / distortion coefficient; when the posture of the fisheye camera changes, a rotation matrix is obtained through calculation according to the real-time posture, a second spherical coordinate table is obtained through reverse rotation calculation of the first spherical coordinate table according to the rotation matrix, and the second spherical coordinate table is converted into a target image coordinate table according to the coordinate conversion relation table; and performing remapping processing on the obtained original fisheye image according to the target image coordinate table to obtain a target fisheye image. According to the invention, distortion correction can be carried out on the fisheye image in real time according to the attitude, and the image quality of the fisheye image is improved.

The invention discloses a pipe wall fisheye distortion correction method based on a multi-scale discriminator and center self-attention, the method is used for pipe wall fisheye distortion network correction, a distortion network is composed of a generator and a discriminator, the generator comprises a flow estimation module and a distortion correction module, and the flow estimation module and the distortion correction module both adopt UNet structures. After down-sampling and up-sampling are carried out on an input image for multiple times, feature reconstruction is carried out to obtain a corrected image, and a discriminator is used for discriminating; the discriminator adopts multi-scale convolution to extract features, and introduces a center self-attention module to highlight center area features. Aiming at the characteristic that fisheye image information is concentrated in a central area, a central self-attention module is introduced into a generator and a discriminator, so that the quality and the structural similarity of a generated image and the performance of network extraction features are improved; a multi-scale discriminator network is constructed, rich image features are obtained, the performance of the discriminator network is improved, and the whole network is guided to generate images with higher quality.