145 results about "Bundle adjustment" patented technology

The invention relates to an unmanned aerial vehicle aerial photography sequence image-based slope three-dimension reconstruction method. The method includes the following steps that: feature region matching and feature point pair extraction are performed on un-calibrated unmanned aerial vehicle multi-view aerial photography sequence images through adopting a feature matching-based algorithm; the geometric structure of a slope and the motion parameters of a camera are calculated through adopting bundle adjustment structure from motion and based on disorder matching feature points, and therefore, a sparse slope three-dimensional point cloud model can be obtained; the sparse slope three-dimensional point cloud model is processed through adopting a patch-based multi-view stereo vision algorithm, so that the sparse slope three-dimensional point cloud model can be diffused to a dense slope three-dimensional point cloud model; and the surface mesh of the slope is reconstructed through adopting Poisson reconstruction algorithm, and the texture information of the surface of the slop is mapped onto a mesh model, and therefore, a vivid three-dimensional slope model with high resolution can be constructed. The unmanned aerial vehicle aerial photography sequence image-based slope three-dimension reconstruction method of the invention has the advantages of low cost, flexibility, portability, high imaging resolution, short operating period, suitability for survey of high-risk areas and the like. With the method adopted, the application of low-altitude photogrammetry and computer vision technology to the geological engineering disaster prevention and reduction field can be greatly prompted.

A method and apparatus for recovering a three-dimensional (3D) scene from two-dimensional (2D) images. A sequence of images is divided into a number of smaller segments and a 3D reconstruction is performed on each segment individually. All the reconstructed segments are then combined together through an efficient bundle adjustment to complete the 3D reconstruction. Segmenting may be achieved by dividing the segments based on the number of feature points that are in each frame. The number of frames per segment is reduced by creating virtual key frames. The virtual key frames encode the 3D structure for each segment, but are only a small subset of the original frames in the segment. A final bundle adjustment is performed on the virtual key frames, rather than all of the original frames. Thus, the final bundle adjustment is two orders of magnitude faster than a conventional bundle adjustment.

To realize high speed and high precision with device and method of three-dimensional measurement by applying estimating process to points corresponding to feature points in a plurality of motion frame images. With the device and method of calculating location information through processes of choosing a stereo pair, relative orientation, and bundle adjustment and using corresponding points of feature points extracted from respective motion frame images, each process is made up of two stages. To the first process section (stages:5A and 6A), robust estimation is applied to determine estimated values of calculation parameters. In the second process section (stages:5B and 6B), corresponding points and stereo pair of large residual errors are excluded, and estimating calculation is carried out again to make measurements in three dimensions.

A method for generating geographic maps for large-range complex urban environment includes extracting two-dimensional roads and two-dimensional building contour information, generating a two-dimensional geographic map; a monocular PTZ camera, under the guidance of the two-dimensional geographic map, conducting video acquisition and obtaining video images; extracting key frames from the video images, extracting line segment characteristics of matching adjacent key frames to obtain line segment correspondence, positioning and establishing an environment three-dimensional geographic model based on a mobile robot of matching line characteristics; matching a three dimensional geographic model and the two-dimensional geographic map, determining the corresponding relation between the two, and generating an enhanced city geographic map; and optimizing the enhanced city geographic map based on bundle adjustment algorithm with geometrical constraints. Compared with the prior art, the method is wide in application scope, high in robustness and precision, small in storage space needed, and low in cost.

Novel tools and techniques for defining a search range for conjugate points in a set of images. In one technique, an intra-overlap area in two images can be used to define a search range in which a conjugate point can be found; in an aspect, this search range might be expressed as a distance range from a station at which one or both of the two images were captured. That distance range can be used to narrow the search range in an image captured from another station, substantially reducing both the computation time to identify a conjugate match and the likelihood of identifying an incorrect match.

The invention discloses a high-precision registration method for a ground laser radar and a close-range photography measurement data. The method comprises the following steps: obtaining multiple close-range photographic images of a scanned object and obtaining a three-dimensional point cloud of the scanned object by a ground laser radar; carrying out SIFT feature extraction and matching based on the pyramid image of the multiple close-range photographic images and generating a free network model of the scanned object according to the matched result; carrying out bundle adjustment on the free network model by using feature points of the three-dimensional point cloud and the image as control points so as to realize coarse positioning on the multiple close-range photographic images; on the basis of a three-dimensional point cloud depth image model generated by point cloud sampling based on the uniform grid of the reference surface of the three-dimensional point cloud, completing prediction and dense matching of corresponding image points of two adjacent close-range photographic images among the multiple close-range photographic images and generating an image point cloud of the scanned object; and carrying out ICP loop iteration registration on the obtained three-dimensional point cloud and image point cloud of the scanned object to realize high-precision registration.

The invention discloses a method for improving camera calibration accuracy by using a multi-plane calibration board. The method comprises the following steps: splicing two calibration substrates drawnwith checkerboard dot patterns to obtain a calibration board; calculating three-dimensional world coordinates of chessboard corners, and merging the three-dimensional world coordinates in the same three-dimensional world coordinate system; acquiring an image of the calibration plate by using a to-be-calibrated camera, and performing filtering preprocessing to obtain a calibration image; determining target corner points on the calibration image and obtaining image pixel coordinates, wherein the image pixel coordinates of the target corner points are in one-to-one correspondence with the three-dimensional world coordinates of the chessboard corner points; calculating and solving a reprojection transformation equation, and performing decomposing to obtain a rotation vector and a translationmatrix; converting the rotation vector into a corresponding rotation matrix, and optimizing the rotation matrix and the translation matrix through a bundle adjustment method after reprojection. The multi-plane calibration plate is formed by splicing the two calibration substrates, the camera calibration precision can be improved, and the problem that errors are large in camera pose estimation is effectively solved.

The invention discloses a systematic geometric demarcation method for reflection three-dimensional measurement of a stripe, and relates to systematic geometric demarcation during the three-dimensional appearance measurement of an object with a mirror reflection characteristic. As a stripe display device appears in a viewing field range of a video camera indirectly, the conventional systematic geometric demarcation process is fulfilled by adopting a method of sticking a marking point on a plane mirror. However, the physical coordinates of the marking point are required to be pre-measured through other precision measurement means. To solve the problem, the systematic geometric demarcation method adopts the scheme that the operation of sticking the marking point on the plane mirror is eliminated, the stripe on the stripe display device is reflected three times by the plane mirror, and analysis and linear calculation are conducted on the coordinates of the characteristic point and the imaging point of the stripe, so that the original value evaluation of the systematic geometric demarcation process is fulfilled; and then the evaluation result is optimized by adopting bundle adjustment, so as to obtain the final systematic geometric demarcation result. The systematic geometric demarcation method has the advantages of simplicity and flexibility; and only a plane mirror with a proper size is required for fulfilling the systematic geometric demarcation of the reflection three-dimensional measurement of the stripe. The systematic geometric demarcation method provides an efficient systematic demarcation way for phase position measurement deflection methods, stripe reflection photogrammetric survey and like which are based on stripe reflection three-dimensional measuring methods, thereby having a wide application prospect.

Novel tools and techniques that can be used to enhance the effectiveness of photogrammetric tools, such as bundle adjustment. One set of techniques can include a photo-observable backsight operation, in which a position of target point can be observed non-photographically, and this observed position can be used to constrain a pixel coordinate location of the same target point in a bundle adjustment operation. Using another technique, a photo-observable check shot operation, the observed position of another target point can be used to verify the validity of a bundle adjustment calculation. Such techniques can be used together or separately.

The invention relates to a tightly coupled binocular vision-inertial SLAM method using combined point-line features, comprising the following steps: determining a transformation relationship between acoordinate system of a camera and a coordinate system of an inertial measurement unit (IMU); establishing a thread of point-line features and IMU tracking to solve an initial three-dimensional point-line coordinate; using the IMU to predict the position of the point-line features to correctly establish associations between the features and data, and combining a re-projection error of the IMU andthe point-line features to solve the pose transformation of consecutive frames after initializing the IMU; establishing a thread of the local bundle adjustment of the point-line features and the IMU,optimizing the three-dimensional point-line coordinate, the pose of a key frame and a state quantity of the IMU in a local key frame window; and establishing a loop back detection thread for the point-line features, using the point-line features to weight and calculate the score of a word bag model to detect the loop back, and optimizing the global state quantity. The tightly coupled binocular vision-inertial SLAM method using the combined point-line features is capable of ensuring stability and high precision in the case where the number of feature points is few and the camera is moving quickly.

The invention relates to a three-dimensional reconstruction method based on bundle adjustment. The three-dimensional reconstruction method comprises the following steps of: calibrating parameters of the interior orientation of a digital camera in a control field through a first iterative process; calibrating parameters of the exterior orientation of the digital camera through a space resection method with a second iterative process in a measuring field by using the parameters of the interior orientation as a given value; and acquiring three-dimensional spatial values of a plurality of measuring points through a bundle adjustment method based on a collinearity condition equation by using the parameters of the interior orientation and the parameters of the exterior orientation as initial values and using image point coordinates of a plurality of control points and a plurality of measuring points in the measuring field as observed values. By the three-dimensional reconstruction method based on the bundle adjustment in an embodiment of the invention, the positioning accuracy of spatial measuring points can be improved in a three-dimensional reconstruction process.

The invention discloses a simultaneous positioning and dense three-dimensional reconstruction method, and the method can process quick motion and frequent closing of a loop in a robust manner, and can operate all the time in a larger-scale scene. The method not only can be used for scanning a high-quality three-dimensional model, but also can meet the demands of virtual reality and augmented reality. The method proposes a simultaneous positioning and map construction framework based on a key frame. At first, the method employs the depth and color information at the same time, and can operate on a CPU at a high speed based on the positioning of the key frame, and also can operate in a challenging scene in a robust manner. In order to reduce an accumulated error, the method introduces increment bundling adjustment, can greatly reduce the calculation amount, and enables the local and global bundling adjustment to be completed in a unified framework. Secondly, the invention proposes a fusion method based on the key frame, and the method can achieve the online generation of a model and updates the three-dimensional model in real time during the adjustment of the posture of the key frame.

The invention protects a monocular vision three-dimensional environment reconstruction method based on Harris-SIFT-BRIEF algorithms; the method comprises the following steps: firstly, calibrating a camera, and solving camera internal and external parameters; then, obtaining video data, using a program to process the data, and thus obtaining an image sequence; combining the Harris algorithm, the SIFT algorithm, and the BRIEF algorithm to extract image characteristic points, and using an optical flow method to match the extracted characteristic points; then, aiming at the obtained plurality of matched point pair sets, combining the calibrated internal and external parameters, using a triangle measuring principle to calculate three dimensional coordinates, obtaining reconstruction scales according to the camera height from the ground or other sensors, thus reconstructing environment three dimensional information, and using a global bundle adjustment and local bundle adjustment combined method to optimize the information; finally, sending reconstruction three dimensional point cloud barrier information to a decision system, and allowing the decision system to decide the steering wheel to turn and control the accelerator. The method can effectively extract image information, thus providing more accurate matching results.

The invention brings forward an aerial photography image and geographical data splicing method based on ORB feature matching. The method comprises the following steps: step 1, reading images to be spliced and geographical information; step 2, distributing tasks to processors; step 3, extracting ORB features of the images to be spliced; step 4, performing initialization estimation on parameters of a camera shooting the images to be spliced and solving a rotation matrix; step 5, improving estimation precision by use of bundle adjustment; step 6, performing initial splicing on the images to be spliced; step 7, performing brightness increment compensation on the images after the splicing, and based on multiband fusion of an image pyramid, performing conditional interpolation fusion on the geographical information after the splicing to obtain splicing result images; and step 8, and performing primary integrated splicing fusion on the splicing result images processed through all threads of each processor, and performing merging to obtain a final result image.

To realize high speed and high precision with device and method of three-dimensional measurement by applying estimating process to points corresponding to feature points in a plurality of motion frame images. With the device and method of calculating location information through processes of choosing a stereo pair, relative orientation, and bundle adjustment and using corresponding points of feature points extracted from respective motion frame images, each process is made up of two stages. To the first process section (stages: 5A and 6A), robust estimation is applied to determine estimated values of calculation parameters. In the second process section (stages: 5B and 6B), corresponding points and stereo pair of large residual errors are excluded, and estimating calculation is carried out again to make measurements in three dimensions.

The invention provides an oppositely-closing attitude-adjusting method based on a multi-laser tracker measuring field. The oppositely-closing attitude-adjusting method comprises the steps that S1, anideal assembly model is established; S2, the multi-laser tracker measuring field is constructed; S3, a bundle adjustment method is adopted for iterative computation, and a homogeneous transformation matrix between any two laser trackers is solved; S4, coordinates of all enhanced reference systems (ERS) under an overall coordination system are calculated; and S5, target workpiece tools are assembled onto standard workpiece tools. Position selection of the ERSs is small in constraint and more flexible, drifting of the relative positions between all the ERSs is avoided, and the measuring precision is improved. Station transferring computation adopts the bundle adjustment method based on an overall optimization idea, station transferring computation between the multiple laser trackers can be completed at a time, and the station transferring precision is improved. Meanwhile, actually measured data of the laser trackers are adopted to guide attitude adjusting, the influence of placing errorof target measuring points on the position estimation result of the target measuring points is reduced, and the measuring precision is improved.

The invention belongs to the technical field of mine production management, and particularly relates to an open-pit mine mining and stripping quantity calculating method utilizing plane image aerial-photographing. The method includes utilizing plane image aerial-photographing technology; performing plane flight path planning and designing according to geographic conditions of an open-pit mine; checking a digital camera carried by a plane; controlling the digital camera for aerial photographing and image acquiring and storing at the same time; recording plane position and posture information provided during photographing; building an open-pit mine earth surface three-dimensional point cloud model. The open-pit mine earth surface three-dimensional point cloud model can be built without adopting a bundle adjustment method of control points, an open-pit mine earth surface irregular triangular net model is built, mining quantity and rock discharging quantity are automatically measured and recorded, accumulated volume values of the open-pit mine earth surface triangular net model at different times are subtracted to accurately calculate earth-rock quantity of an excavating or backfilling body, calculating of open-pit mine mining and stripping quantity is realized, open-pit mine measuring work efficiency is improved, and statistical accuracy of mining and stripping quantity is improved.

The invention discloses a mine multi-camera video fusion method based on GPU and CPU cooperative computing and belongs to the mine multi-camera video fusion method. The method is characterized in that a GPU and CPU cooperative computing model is established, and thus current situation of poor real-time performance of an existing video fusion method is relieved; automatic matching is carried out by utilizing features of overlapped regions between adjacent video sources to determine relative position relation of the video sources, so that positions of cameras do not need to be assigned in advance, and automatic integration is realized; camera parameter bundle adjustment estimation and automatic alignment are carried out, so that the problem of deformation of panorama images after fusion is solved; and an exposure compensation and multiband fusion method is adopted, and finally, an ultrahigh-resolution large-visual angle mine mining (extraction) working face large-scale scene real-time integral video is generated, so that the method plays an important role for improving coal mining work efficiency and guaranteeing coal mine safety production. The advantages are that the multi-camera video fusion method utilizes the existing cameras of a coal mine fully to generate the mine mining (extraction) working face panorama video, thereby reducing equipment investment and improving safety of coal mine production.

The invention relates to a building elevation positioning control and precision evaluation method based on satellite images, belonging to the fields of acquisition and updating of urban spatial data of satellite three-dimensional remote sensing data. The method comprises the following concrete steps: (1) carrying out the field survey by a GPS to acquire ground control point data; (2) carrying out block bundle adjustment on aerial images to acquire building roof coordinates; (3) integrating the acquired horizontal coordinate with the building roof vertical coordinate acquired by airborne LiDAR data to acquire a building roof control point; (4) geometrically positioning high-resolution satellite three-dimensional images based on a rational function model (RFM) in an image space to acquire three-dimensional coordinates of the ground and the building roof; and (5) acquiring the relation between the elevation of the control point and the elevation precision of the point pattern of a check point by adopting the building elevation positioning control and precision evaluation scheme based on the point pattern. The invention can promote the application of high-resolution satellite three-dimensional remote sensing images in the fields of acquisition and updating of urban spatial data.

The invention discloses a strict calibration method of panorama camera, mainly comprising the following steps: building a panorama camera calibration field, and measuring object space coordinate of a control point; obtaining multiple groups of single camera calibration images of the panorama camera calibration field according to photographic parameters set in advance; measuring image point coordinate of a connection point between the control point and the image on the image of each single camera; carrying out bundle adjustment on strip block formed by images of each single camera, obtaining exterior orientation element of single camera image of each exposure point position in an object space control point coordinate system; carrying out relative orientation between single camera images for calibration image of one exposure point, and obtaining a relative orientation element initial value; and according to the two initial resolving results above, placing the single camera images of all exposure points together to form an integer, building error equation, executing strict adjustment treatment, and thereby obtaining strict relative orientation element between single cameras in the panorama camera. The method provided by the invention can effectively improve calibration precision of the panorama camera.

The present invention provides a method for implementing high-precision orientation and evaluating orientation precision of a large-scale dynamic photogrammetry system, including steps: a) selecting a scale bar, arranging code points at two ends of the scale bar, and performing length measurement on the scale bar; b) evenly dividing a measurement space into multiple regions, sequentially placing the scale bar in each region, and photographing the scale bar by using left and right cameras; d) limiting self-calibration bundle adjustment by using multiple length constraints, adjustment parameters including principal point, principal distance, radial distortion, eccentric distortion, in-plane distortion, exterior orientation parameter and spatial point coordinate; and e) performing traceable evaluation of orientation precision of the photogrammetry system. The present invention can effectively reduce the relative error in length measurement.

The invention relates to a method for aerophotogrammetric bundle adjustment based on parameterization of polar coordinates. The method comprises the following steps of: 1) acquiring a series of images through an aerophotogrammetric area, and extracting and matching characteristic points of all images in the aerophotogrammetric area; 2) expressing the characteristic points based on parameterization of the polar coordinates; 3) establishing an observation equation of the bundle adjustment based on the parameterization of the polar coordinates; and 4) carrying out the adjustment for an area network with the observation equation based on the parameterization of the polar coordinates. According to the method, the traditional form of the parameterization of rectangular coordinates XYZ for expressing three-dimensional characteristic points is changed and replaced by expression with main camera points, observation vectors and polar coordinate quantities in the polar coordinates more close to a measurement space of a camera, and the method can be widely applied to aerial triangulation in aerophotogrammetry.

The invention discloses a rapid three-dimensional scanning method for a human foot, and relates to the technical field of three-dimensional scanning and digital analysis. The method includes the steps: calibrating a measuring head by means of internal and external parameter bundle adjustment; projecting speckles to the foot, acquiring a foot image and matching the speckles; performing three-dimensional density point cloud reconstruction by the calibrated measuring head to obtain a three-dimensional foot model; extracting the size of the foot model according to the three-dimensional foot model.According to the rapid three-dimensional scanning method for the human foot, coordinates of foot key points such as an inner ankle point, an outer ankle point, a heel point and a tiptoe point of thehuman foot can be acquired, sizes such as foot sweep, the length of the foot, the length of an outer ankle, foot breadth and heel breadth are automatically extracted and calculated, and scanning speedand precision are high.

The invention provides a multi-scale point cloud noise detection method based on density analysis. The method mainly comprises the steps that firstly, a multi-scale density analysis algorithm is utilized for preliminarily judging points which possibly are noise; secondly, triangulation network constrains are utilized for classifying the points which are detected as noise in the last step by mistake to be non-noise points again. The multi-scale point cloud noise detection method based on density analysis can effectively detect out isolated noise and cluster noise contained in point cloud of laser radar and contained in point cloud obtained through image matching, and has the good application prospect in the fields of laser radar point cloud filter and image matching mistake-matching point detection, bundle adjustment noise detection and the like.

The invention discloses an image depth estimation method and system based on a CNN (Convolutional Neural Network) and a depth filter. The image depth estimation method comprises the steps of obtainingimage depth estimation values based on the CNN, extracting local features points from the image, and establishing a minimum photometric error equation to solve the relative pose of a camera; optimizing the camera pose based on the feature points matched by polar line searching, selecting an image block by taking the feature points as a center, performing polar line searching on the image block toobtain the optimum matching, constructing a bundle adjustment equation according to the optimum matching to optimize the relative pose; and performing depth value filtering based on the camera pose,namely, filtering depth values by adopting Gaussian integration until the depth values converge. The image depth estimation method overcomes a problem of inaccurate image depth and a problem of absolute scale loss of monocular vision, and can be applied to many fields such as three-dimensional scene reconstruction, indoor positioning and augmented reality.

The invention discloses a multi-camera movable photography measuring method in a vibration environment, comprising the following steps: printing a coding marker on a measured object or nearby rigid body in a sticky mode, then mounting two digital cameras in rigid connection at each measuring point (one mounted horizontally and the other mounted vertically) to obtain exterior orientation elements of two digital cameras in the measuring points, and then accurately gaining a three-dimensional coordinate of the printed markers on the measured object in the vibration environment through a bundle adjustment computing method. By improving the computing method, the three-dimensional coordinate of the printed markers on the measured object in the vibration environment is gained accurately through the bundle adjustment computing method, rather than a compensation algorithm. Three-dimensional measurement problems of position, shape or movement of the object in the vibration environment are solved, such as accurate measurement of model deformation in a wind tunnel test.

The invention provides an SLAM method based on a semantic bundle adjustment method, and belongs to the field of mobile robot simultaneous localization and mapping (SLAM). The method is characterized in that the method combines a 6DOF object and the camera attitude through new semantic global optimization, and can work under a 2D or 3D sensor. As semantic information is added, a target detection channel can be seamlessly integrated into a BA type optimization system of an SLAM system based on BA without peripherals. The method of the invention is simple and easy to implement, and has strong practicability. SLAM constraints can be used in robust target detection, and can adapt to a more complicated environment.

The invention discloses a high-precision real-time stereoscopic visual positioning method utilizing the parallax space bundle adjustment. According to the invention, only a video streaming obtained by a stereo camera is utilized to carry out navigation positioning of a mobile robot, so that the high-precision real-time stereoscopic visual positioning method has complete autonomy; a U-SURF descriptor is utilized to carry out feature matching, so that the high-precision real-time stereoscopic visual positioning method has strong robustness on illumination variation, motion blur and large rotation and scale variation which are generated in the moving process; and due to the utilization of a parallax space bundle adjustment optimizing method, an autonomous navigation vehicle can also be subjected to accurate real-time positioning when traveling for a long distance, so that the high-precision real-time stereoscopic visual positioning method has the advantages of low degree of dependence on an optimal initial value, high convergence rate, high value stability and the like and compared with a stereoscopic visual positioning method utilizing the bundle adjustment, the high-precision real-time stereoscopic visual positioning method has higher positioning precision and higher speed. The integral stereoscopic visual positioning method disclosed by the invention has high calculating speed and can be operated on line in real time.