1288 results about "Monocular camera" patented technology

The present invention provides a system (method and apparatus) for creating photorealistic 3D models of environments and/or objects from a plurality of stereo images obtained from a mobile stereo camera and optional monocular cameras. The cameras may be handheld, mounted on a mobile platform, manipulator or a positioning device. The system automatically detects and tracks features in image sequences and self-references the stereo camera in 6 degrees of freedom by matching the features to a database to track the camera motion, while building the database simultaneously. A motion estimate may be also provided from external sensors and fused with the motion computed from the images. Individual stereo pairs are processed to compute dense 3D data representing the scene and are transformed, using the estimated camera motion, into a common reference and fused together. The resulting 3D data is represented as point clouds, surfaces, or volumes. The present invention also provides a system (method and apparatus) for enhancing 3D models of environments or objects by registering information from additional sensors to improve model fidelity or to augment it with supplementary information by using a light pattern projector. The present invention also provides a system (method and apparatus) for generating photo-realistic 3D models of underground environments such as tunnels, mines, voids and caves, including automatic registration of the 3D models with pre-existing underground maps.

The invention discloses a picking method and picking robot device aiming at fruits which are in size of an apple and is similar to a sphere. The picking robot device comprises a mechanical actuating device, control system hardware and control system software. The mechanical actuating device comprises a picking mechanical arm, an underactuated manipulator, an electric sliding table and an intelligent mobile platform, wherein the control system hardware comprises an IPC (industrial personal computer), a motion control card, a data acquisition card, an AHRS (attitude and heading reference system), a coder, a monocular camera, a binocular camera, a force sensor, a slipping sensor and the like. During operation, the IPC fuses information of the coder, the AHRS, monocular camera components and an ultrasonic sensor to enable the mobile platform to independently navigate and avoid obstacles. A binocular vision system collects images of mature fruits and obstacles and extracts the characteristics of the images so as to realize obstacle avoidance of the mechanical arm and fruit positioning. Finally, the IPC fuses the information of the force sensor, the slipping sensor and the position sensor, thereby further reliably gripping the mature fruits and separating the fruits from fruit branches.

The invention discloses a binocular stereo vision three-dimensional measurement method based on line structured light scanning, which comprises the steps of performing stereo calibration on binocularindustrial cameras, projecting laser light bars by using a line laser, respectively acquiring left and right laser light bar images, extracting light bar center coordinates with sub-pixel accuracy based on a Hessian matrix method, performing light bar matching according to an epipolar constraint principle, and calculating a laser plane equation; secondly, acquiring a line laser scanning image of aworkpiece to be measured, extracting coordinates of the image of the workpiece to be measured, calculating world coordinates of the workpiece to be measured by combining binocular camera calibrationparameters and the laser plane equation, and recovering the three-dimensional surface topography of the workpiece to be measured. Compared with a common three-dimensional measurement system combininga monocular camera and line structured light, the binocular stereo vision three-dimensional measurement method avoids complicated laser plane calibration. Compared with the traditional stereo vision method, the binocular stereo vision three-dimensional measurement method reduces the difficulty of stereo matching in binocular stereo vision while ensuring the measurement accuracy, and improves the robustness and the usability of a visual three-dimensional measurement system.

The invention discloses a three-dimensional human body posture estimation method and device. The method comprises the following steps of S1, using a monocular camera to collect the depth images and the RGB color images of a human body at different angles; S2, constructing a human body skeleton key point detection nerve network based on the RGB color images, and obtaining a key point labeling image; S3, constructing hand joint node 2D-3D mapping network; S4, calibrating the depth image of the human body at the same angle and the key point labeling image, and then carrying out three-dimensionalpoint cloud color conversion on the corresponding depth image so as to obtain a colored depth image; S5, based on the key point labeling image and the colored depth image, using a preset learning network to predict the corresponding position of the labeled human body skeleton key point in the depth image; and S6, combining the outputs of the step S3 and the step S5 to achieve the refined estimation of three-dimensional human body posture estimation.

The invention discloses a vehicle ranging method based on monocular vision, and belongs to the field of target detection and ranging. The method comprises the steps of installing a monocular camera on a vehicle, measuring the height and the pitching angle of the camera, and determining focal distance parameters of the camera; acquiring a video image in an expressway environment; performing preliminary de-noising and filtration on the video image by adopting Gaussian filtration; performing interest region segmentation preprocessing before detection of a target vehicle on the video image; performing vehicle detection in a segmented video image region, wherein an Haar feature for increasing a wheel feature and a tail feature is adopted in the target vehicle detection, so as to effectively improve the accuracy of target vehicle identification; and measuring the distance of the target vehicle, wherein a ranging method based on pin-hole imaging is adopted within a short-distance range, while a ranging method of data fitting is adopted within a long-distance range, so that the error rate is reduced, and a real-time ranging effect can be achieved. The method has the advantages of high detection speed, high accuracy, strong real-time property and low cost.

The invention discloses an image appearance based loop closure detecting method in monocular vision SLAM (simultaneous localization and mapping). The image appearance based loop closure detecting method includes acquiring images of the current scene by a monocular camera carried by a mobile robot during advancing, and extracting characteristics of bag of visual words of the images of the current scene; preprocessing the images by details of measuring similarities of the images according to inner products of image weight vectors and rejecting the current image highly similar to a previous history image; updating posterior probability in a loop closure hypothetical state by a Bayesian filter process to carry out loop closure detection so as to judge whether the current image is subjected to loop closure or not; and verifying loop closure detection results obtained in the previous step by an image reverse retrieval process. Further, in a process of establishing a visual dictionary, the quantity of clustering categories is regulated dynamically according to TSC (tightness and separation criterion) values which serve as an evaluation criterion for clustering results. Compared with the prior art, the loop closure detecting method has the advantages of high instantaneity and detection precision.

The invention provides a simultaneous localization and mapping (SLAM) method based on parking locations and two-dimensional codes. A surround view system based on a low-cost front-view monocular camera and composed of a fisheye camera performs robust identification of parking locations, parking location numbers and pre-arranged two-dimensional codes in an environment; the heading and speed of a vehicle are observed by using a low-cost inertial navigation device; and the optimal estimation of the parking locations and marking locations of the two-dimensional codes is realized by an SLAM algorithm, a garage plane distribution map is automatically created, and the real-time and high-precision decimeter localization of an unmanned vehicle is realized by using the mapping result.

The invention discloses a monocular vision and quick-response code road sign based indoor autonomous navigation method. The method includes the steps of S1), acquiring an initial position of a robot, and setting a quick-response route according to the initial position and a target position of the robot; S2, detecting a quick-response code in real time by a monocular camera of the robot, calculating relative position relation among the robot and the quick-response code according to internal and external parameters and determining current navigation information of the robot; S3, performing motion according to the current navigation information of the robot, judging whether a destination address is the target position or not, if yes, completing navigation, or otherwise, performing angle adjustment according to the navigation direction of the quick-response code and turning the step S2). With the method, the problem about complicated calculation of the monocular camera can be solved, and the requirement on accuracy of indoor autonomous navigation can be met.

The invention belongs to the field of intelligent unmanned intelligent ships, and relates to an unmanned ship water surface target detection, identification and positioning method based on monocular camera and lidar information fusion. The detection, identification and positioning of a water surface target by an unmanned ship are influenced by distances and the fluctuation of the target, so that alidar and a monocular camera are integrated to accurately detect, identify and position the target within a sensing range. According to the method, acquired water surface target images are adopted totrain a neural network-based target detection and recognition model; the lidar employs a conditional removal filter and Euclidean clustering to obtain the position of the water surface target in a world coordinate system; and finally, a camera image information and lidar information fusion method is designed, and therefore, the method is highly robust to uncertain factors. With the method adopted, the unmanned ship is capable of accurately detecting, identifying and positioning the water surface target; and good environment perception can be realized for the target tracking, path planning andautonomous navigation of the unmanned ship. The method has a broad application prospect.

The invention relates to a method for real-time vision positioning for a monocular camera, which belongs to the field of computer vision. The method comprises the following steps: firstly, acquiring an object image characteristic point set to establish an object image database and perform real-time trainings; secondly, modeling the camera to acquire model parameters of the camera; and thirdly, extracting a real-time image characteristic point set by the camera, matching real-time image characteristic points with the characteristic point set in the object database, and removing error matching and performing an affine inspection to acquire characteristic point pairs and object type information. The characteristic point pairs and the object type information are used to perform characteristic tracking, and accurate tracking characteristic points of the object image are combined with the model parameters of the camera so as to acquire three-dimensional poses of the camera. The method can achieve the functions of self-positioning and navigation by using a single camera only, thus the system complexity and the cost are reduced.

An embodiment of the present invention includes: a tracking object image extracting section that extracts a tracking object image, which represents a tracking object, from an image captured by a monocular camera; a two-dimensional displacement calculating section that calculates, as actual movement amounts, amounts of inter-frame movement of the tracking object image; a two-dimensional plane projecting section that generates on a two-dimensional plane a projected image of a three-dimensional model, which represents in three dimensions a capturing object captured by the monocular camera; a small motion generating section that calculates, as estimated movement amounts, amounts of inter-frame movement of the projected image; and a three-dimensional displacement estimating section that estimates amounts of three-dimensional motion of the tracking object on the basis of the actual movement amounts and the estimated movement amounts.

An apparatus discriminates a potential obstacle in the path of a vehicle from among various objects in an image shot by a monocular camera. First, an object detecting unit detects an object in the image by applying a saliency calculation to the image. Second, an object discriminating unit discriminates an object from among the objects detected by the object detecting unit as a potential obstacle by applying a neural network method to the objects.

The invention discloses a simultaneous localization and mapping (SLAM) method for an unmanned aerial vehicle based on mixed vision odometers and a multi-scale map, and belongs to the technical field of autonomous navigation of unmanned aerial vehicles. According to the SLAM method, an overlooking monocular camera, a foresight binocular camera and an airborne computer are carried on an unmanned aerial vehicle platform; the monocular camera is used for the visual odometer based on a direct method, and binocular camera is used for the visual odometer based on feature point method; the mixed visual odometers conduct information fusion on output of the two visual odometers to construct the local map for positioning, and the real-time posture of the unmanned aerial vehicle is obtained; then theposture is fed back to a flight control system to control the position of the unmanned aerial vehicle; and the airborne computer transmits the real-time posture and collected images to a ground station, the ground station plans the flight path in real time according to the constructed global map and sends waypoint information to the unmanned aerial vehicle, and thus autonomous flight of the unmanned aerial vehicle is achieved. Real-time posture estimation and environmental perception of the unmanned aerial vehicle under the non-GPS environment are achieved, and the intelligent level of the unmanned aerial vehicle is greatly increased.

A method for detecting moving objects including people. Enhanced monitoring, safety and security is provided through the use of a monocular camera and a structured light source, by trajectory computation, velocity computation, or counting of people and other objects passing through a laser plane arranged perpendicular to the ground, and which can be setup anywhere near a portal, a hallway or other open area. Enhanced security is provided for portals such as revolving doors, mantraps, swing doors, sliding doors, etc., using the monocular camera and structured light source to detect and, optionally, prevent access violations such as “piggybacking” and “tailgating”.

The invention provides an indoor and independent drone navigation method based on three-dimensional vision SLAM. The indoor and independent drone navigation method comprises the steps that an RGB-D camera obtains a colored image and depth data of a drone surrounding environment; a drone operation system extracts characteristic points; the drone operation system judges whether enough characteristicpoints exist or not, if the quantity of the characteristic points is larger than 30, it shows that enough characteristic points exist, the drone attitude calculation process is conducted, or, relocating is conducted; a bundling optimizing method is used for global optimization; an incremental map is built. Drone attitude information is given with only one RGB-D camera, a three-dimensional surrounding environment is rebuilt, the complex process that a monocular camera solves depth information is avoided, and the problems of complexity and robustness of a matching algorithm in a binocular camera are solved; an iterative nearest-point method is combined with a reprojection error algorithm, so that drone attitude estimation is more accurate; a drone is located and navigated and independentlyflies indoors and in other unknown environments, and the problem that locating cannot be conducted when no GPS signal exists is solved.

The invention discloses an industrial robot capturing method based on a monocular camera and a three-dimensional force sensor. The method comprises the steps that the monocular camera and the industrial robot are calibrated, and the three-dimensional force sensor is initialized; the monocular camera acquires an object image, an upper computer conducts image processing and calculates the position information of an object, the robot is controlled to move to a position above the object according to the position information, and then the robot moves downwards vertically till making contact with the object to be captured; the three-dimensional force sensor senses the contact force information between the tail end of the robot and the surface of the object, the upper computer controls the industrial robot to stop moving according to contact force information and calculates the pose increment of the tail end of the robot, the posture of the robot is adjusted according to the pose increment till a pneumatic sucker at the tail end of the robot is attached to the plane of the object to be captured, and the pneumatic sucker acts to achieve capturing. By the adoption of the method, objects placed obliquely in a messy mode can be captured, and complexity of camera calibration and object space pose estimation is reduced.

The invention relates to a vehicle distance measurement method based on the Monocular Camera Machine Vision. The method is consisted of an image capturing system adopting monocular camera and a multi-function image processing and analyzing system. The method starts from image information of the monocular camera image capturing system to process for the image information, uses an image recognition technology to extract and remedy to gain a size of standard object edge, reverses and rebuilds space information of the standard object according to the size of the standard object accounted in the image and a pre-calibration value, and processes to calculate a relative position of the standard object in the three-dimensional space, so as to realize the vehicle distance measurement. The invention solves problems that the existing monocular camera has a lower recognition rate for the whole vehicle contour and ground road logo line and needs to collect the image of the whole vehicle, has a large calculation, is not applicable to short distance, low speed and large vehicle flow situations.

The invention discloses a method for conducting real-time tracking on faces by a monocular camera. The method comprises the steps that firstly, the camera is opened so as to search for whether the faces exist around the camera or not, images collected by the camera are transmitted to an image processor, the image processor calls an image processing program to conduct image compression, three AdaBoost cascade connection strong classifiers based on Haar characteristics are loaded on the image processing program, skin color detection is carried out on all the compressed images, a detected window is used for detecting the faces with multiple angles through the three AdaBoost cascade connection strong classifiers, after the faces are detected, real-time tracking is carried out on face targets, the difference between the central point coordinates of a face target area and the central point coordinates of the whole images is compared, the camera is adjusted to enable the central point coordinates of the face images to be approximately aligned with central point coordinates of the images, and real-time tracking on the faces is achieved. The method has the advantages that the tracking method is simple, the calculated amount is small in the tracking process, the hardware structure is simple, and real-time tracking of the face targets can be well achieved.

The invention discloses a monocular-camera-based method for measuring depth. In the method, the depth information of a static object is measured rotationally by utilizing a cloud deck. The method comprises the following steps of: collecting an image under the condition of the given azimuth of a camera and the given focal length of a lens; controlling the camera to rotate an angle horizontally or vertically by the cloud deck, and collecting an image again; finding matched image points in the two images by an image matching algorithm; and calculating the depth of object points corresponding to the image points. The invention also discloses a monocular-camera-based method for measuring a depth field, and a system device for measuring in a monocular-camera-based method for measuring the depth and sizes of motion targets. The method is suitable for a large number of ordinary video monitoring cameras with rotary cloud decks which are installed at present, so that the measurement of the depth and actual sizes of objects is realized, and the measurement problem of video monitoring is solved effectively.

The invention discloses a vision navigation method of a mobile robot based on a hand-drawn outline semantic map. The method comprises the following steps: drawing the hand-drawn outline semantic map; selecting a corresponding sub-database; designing and identifying labels; performing object segmentation; matching images included in the sub-database with segmented regions; performing coarse positioning on the robot; and navigating the robot. The unified labels are stuck on possible reference objects in a complex environment, a monocular camera of the robot is utilized as a main sensor for guiding operation of the robot according to guide of the hand-drawn outline semantic map, sonar is utilized for assisting the robot in obstacle avoidance, information of a milemeter is further fused for coarse positioning, and the navigation task is finally completed under mutual coordination of the components. By utilizing the method disclosed by the invention, the robot can realize smooth navigationwithout a precise environment map or a precise operation path and effectively avoid dynamic obstacles in a real-time manner.

The invention discloses a monocular camera and millimeter wave radar fused object identification and positioning method and system, and belongs to the technical field of multi-sensor fusion. The method comprises the following steps: step 1, sensor data acquisition and frame synchronization; 1a, millimeter wave radar and monocular camera data acquisition; 1b, data sequence matching based on frame synchronization; 2, recognition of an object and a boundary based on the neural network processing image; 2a, training in advance an instance partition neural network based on Mask-RCNN; 2b, completingobject recognition and image instance segmentation by using a neural network; 3, coordinate system conversion and data fusion; 3a, under the condition of known external parameters between sensors, converting radar point cloud to an image coordinate system; 3b, completing object matching information fusion based on an instance segmentation result; and 3c, displaying the object category and the object spatial position under the three-dimensional coordinate system. The method is more stable, and the influence of severe weather on data acquisition can be overcome.

The invention discloses a method for selecting an autonomous landing area of an unmanned aerial vehicle (UAV) under a complex environment based on visual SLAM, which is used for solving the technical problem of poor practicability of the existing UAV landing area control method. According to the technical solution, the method comprises the steps of obtaining an image sequence via an overlooking monocular camera carried by a UAV mobile platform, calculating the pose of the UAV in real time via an SLAM algorithm and establishing a sparse point cloud map, and meshing the point cloud map to construct a two-dimensional grid height map; then dividing the grid map according to the height in combination with a Means shift image segmentation algorithm, and finally screening an area that is farthest from a potential obstacle and is suitable for landing of the UVA according to the landing height requirement. According to the method, the pose of the UAV is calculated by adopting the monocular visual SLAM and estimated in real time, the two-dimensional grid height map is constructed, and the area suitable for landing of the UVA is screened. The method, which does not depend on a landmark, has good practicability.