64 results about "Direct linear transformation" patented technology

The invention discloses a multi-rotor unmanned aerial vehicle (UAV) pose acquisition method based on markers in shape of a large and a small square. The method is as below: first determining target markers based on the relative height of the UAV and the markers, wherein the markers comprise a small square and a large square with side length 10 times of that of the small square; second, according to the determined target markers, establishing a relationship model between a CCD camera of the UAV and the target markers; and finally, according to the established model, employing a direct linear transformation method to obtain an azimuth angle of the UAV and location information of the UAV at the corresponding time. The method can accurately obtain the pose of the UAV.

The invention relates to an application of the image mosaic technology in tunnel surrounding rock deformation noncontact measurement, in particular to an application in noncontact measurement of tunnel large-size section surrounding rock deformation. The measurement method includes firstly, arranging points to be measured and image mosaic mark points on a large-size monitoring section, placing portable precise three-dimensional controlling fields in front of the large-size monitoring section, performing convergent photography by an ordinary digital camera to capture monitored section overlapping sub-image groups of the points to be measured, the image mosaic mark points and the controlling fields included during different periods, then carrying out distortion correction and high-precision mosaic to the images to acquire a panorama of the monitoring section; secondly, extracting image point coordinates of the points to be measured and the controlling points in the panorama by using the sub-pixel location technology, analyzing the spatial three-dimensional coordinates of the points to be measured by a direct linear transformation method, calculating the measuring line lengths among the points to be measured during the same period and the changes of the measuring line lengths among the points to be measured during different periods so as to acquire the displacement information. Accordingly, the accurate results of the tunnel large-size section surrounding rock deformation noncontact measurement can be realized.

The invention discloses a monocular vision positioning-based rotor unmanned aerial vehicle attitude estimation method, which comprises the following steps of: firstly, preprocessing an image which iscaptured by an airborne camera and contains a visual mark, eliminating image noise and highlighting the visual mark; According to the geometric characteristics of the visual mark, extracting the visual mark composed of an H-shaped mark and a circular ring through a five-step method; Then, detecting 12 corner points of the H-shaped mark by adopting a distance three-point method, and matching the 12corner points with corresponding corner points in a preset reference image; And finally, calculating a homography matrix according to the matched angular point pairs, decomposing the homography matrix by adopting a direct linear transformation DLT method to calculate an attitude angle of the unmanned aerial vehicle relative to the visual sign, and calculating position data of the unmanned aerialvehicle relative to the visual sign according to a similar triangle principle of camera imaging. The method is higher in calculation speed and has a better practical value.

The embodiment of the invention discloses and provides an infrared image enhancing method. According to the method, images are used for histogram statistics, in addition, the histogram is calculated for obtaining the gray level array of the pixel point number extreme value, the array is used for judging whether the histogram distribution belongs to a single-peak type or a double-peak form or not, the respective unfolding regions and compressions are obtained according to different types of the histogram distribution, the linear conversion is utilized for unfolding and compressing the gray scale level in the unfolding regions and the compression regions by using linear conversion, and finally, the goal of enhancing the images is reached. The algorithm is simple, the complicated algorithm is avoided, and meanwhile, the method overcomes the defects that in the gray level direct linear conversion process, the background and the noise are enhanced, the target gray scale level cannot be effectively enhanced, the image brightness is dark, and the like.

The invention belongs to the field of visual measurement, and relates to a large-view-field camera calibration method based on a laser tracker. According to the method, a calibration board is manufactured according to the characteristics of the laser tracker; the images of the calibration board of different positions and angles are photographed by a camera, and the distribution position of the calibration board covers the field of view of the camera; angular point detection is performed on the calibration board by utilizing angular point detection tool software, and the angular position detection results of the central area of the calibration board are selected; the world coordinate system coordinates of the angular points of the central area of the calibration board are calculated by utilizing the target spherical coordinates of the four corners of the calibration board; and the internal matrix and the external matrix of the camera are calculated by utilizing a direct linear transformation method, and the radial distortion coefficient is iteratively calculated so that a high-precision calibration result is obtained. The problem of the linear transformation method that the large high-precision calibration board requires to be manufactured for the requirement of large-view-field calibration can be solved, and the measurement characteristics of the laser tracker are favorable for arrangement of the calibration board at an appropriate angle so that the operability and the calibration precision can be enhanced. The method is suitable for calibration of the large-view-field camera.

The invention discloses a panoramic image stitching method for a vehicle-mounted environment. On the basis of frequently-used panoramic image stitching, an effective panoramic stitching scheme is provided for panoramic stitching of a vehicle-mounted system. An improved weighted fusion imaging model is used for correcting a fisheye image, a down-looking projection stage uses a direct linear transformation method as a basis, and translation and rotation parameters are added to correct measurement errors in an experiment process; an image registration stage uses an image registration method based on an edge template, arranged white form line four corners serve as a registration template, template edge points need to be selected during the registration process, an edge template descriptor is generated, and thus, the registration is more accurate; an image fusion stage uses a weighted average method as a basis, a trigonometric function term is added, and pixel differences in an overlapped area between adjacent images are reduced. The registration accuracy and the fusion effects are improved, the process is simple, the method is applied to the vehicle-mounted environment, and the application prospect is good.

The invention relates to a three dimensional displacement monitoring method for a strip mine side slope. According to the method, a three dimensional laser scanner is used for obtaining point cloud data of a side slope zone to be monitored of a strip mine, a three dimensional model of the side slope zone to be monitored is built, monitoring images of the side slope zone to be monitored are collected in real time via a monocular camera, and side slope monitoring sequence images can be obtained; control points are selected in a scope, covered by the side slope monitoring sequence images, of the side slope zone to be monitored; a direct linear transformation equation for mutual conversion between geodetic coordinates and image pixel two dimension coordinates is determined; an image local characteristic detecting and coupling algorithm is adopted, and matching characteristic points that are monitored in real time in the side slope monitoring sequence images are extracted, displacement of matching characteristic points in a current image and a previous among the side slope monitoring sequence images can be monitored in real time, and two dimension image displacement of the characteristic points in the side slope monitoring sequence images is converted into three dimension displacement of side slope points in a geodetic coordinate system.

The invention discloses a fluid motion vector estimation method based on a space-time image and belongs to the technical field of flow field measurement. The method firstly adopts a single camera to acquire a fluid image sequence at fixed time intervals; then an aberration correction linear conversion method is adopted to carry out optical distortion correction on images and a group of speed measurement lines are arranged in a flow measurement direction, wherein each speed measurement line uses a pixel as an x-coordinate and a time as a y-coordinate to synthesize a time-space image; then frequency-spectrum analysis is carried out on the time-space images which undergo edge detection and a discrete approximation method is adopted to solve a frequency-spectrum main direction and obtain a texture main direction vertical to the frequency-spectrum main direction; and according to a mapping relation of an object-space coordinate and an image-space coordinate, physical sizes of the speed measurement lines are solved and the value of a one-dimension flow speed vector is calculated through combination of the texture main direction and the time intervals so as to realize time-averaged flow field calibrating. The fluid motion vector estimation method based on the space-time image can be used for time-averaged flow field measurement of natural tracing water flows such as river surfaces and the like and is especially suitable for online monitoring of flow rates of rivers in a flood period.

A traffic accident site rapid mapping system based on image guiding comprises a homography matrix solution module, a space coordinate transformation module, a re-sampling module, a mapping and marking module and a file preserving and printing module. The homography matrix solution module builds a mapping relationship between a traffic accident site picture image plane and an actual road plane according to the two-dimensional direct linear transformation method, the space coordinate transformation module restores the position information of a traffic site picture in object space according to the homography matrix and scales the range of the traffic accident site to the size range of the original traffic accident site picture, the re-sampling module restores gray information of the traffic accident site picture in object space and achieves transformation from a perspective picture to an orthographic projection picture of the traffic accident site picture, the mapping and marking module achieves direct mapping and relevant size marking of the traffic accident site information on the orthographic projection picture, and the file preserving and printing module achieves functions of preservation and site printout of the traffic accident site picture. By means of the traffic accident site rapid mapping system, the accuracy of traffic accident site survey is improved, the on-site data processing and mapping time is saved, and the requirement of site picture printout on the traffic accident site is further satisfied.

The invention discloses a river surface photographic surveying method based on height-change homography. According to the method, a camera is erected on one side of a river bank to shoot a river surface at a gradient visual angle; firstly, 8 non-coplanar square opposite angle markers are uniformly and symmetrically distributed on the two sides of the river to serve as ground control points, the crossing points of marker posts and the water surface and the crossing point of a water gauge and the water surface are used as water control points, and a non-prism total station is adopted for measuring the world coordinates of the water control points; then the sub pixel image coordinates of the ground control points are attracted through a man-machine interaction way, and a DLT (direct linear transformation) factor and an optimized aberration coefficient are solved through direct linear transformation for aberration correction; next, water gauge image reading and water level value attracting are performed automatically through an image identification method, and the gradient coefficient and the water level coefficient are calculated by using the world coordinates of the water control points, so as to build a water surface elevation model; finally, water surface image point coordinates are converted to be object space coordinates by using a height-change homography relation, so as to accomplish fiver photographic surveying. According to the invention, the system distribution is simple, the cost is low, and the method is applicable to high-accuracy river surface photographic speed surveying.

The invention relates to a rapid extraction method based on a self-luminous ball camera calibration target, and belongs to the field of visual measurement. Before calibration, the three-dimensional calibration target is mounted, a three-coordinate measuring machine is used for measurement, accurate 3D position information of the center of the self-luminous ball target is determined, left and right cameras in a binocular visual system are used to shoot the self-luminous ball target, characteristic information of the center of the self-luminous ball target is extracted from pictures, and a 2D pixel coordinate of the center of the self-luminous ball target is calculated. On the basis of a direct linear transformation method, internal and external parameters of the left and right cameras are calculated, and stereoscopic calibration of the cameras in the 3D space is realized. The self-luminous ball targets including different numbers of rows are different in brightness, different grayscale thresholds are set, and thus, space information is identified rapidly and automatically, and the left and right cameras are calibrated rapidly in high precision in the calibration process.

The invention relates to a method for measuring the height of a building on the basis of a single digital photo, and provides a principle and the method for measuring the height of the building by utilization of the single digital photo. The method is widely applied to digital urban construction, and belongs to the field of photogrametry and computer vision. The method is a method for measuring the height of the building by adopting an inverse direct linear transformation computing method on the basis of a direct linear transformation formula, and comprises the following steps: firstly, shooting the digital photo of the building; secondly, measuring image coordinates (u, v) of various points on a height line of the building on the digital photo; thirdly, transforming the image coordinates (u, v) of the various points into a right-handed system image plane coordinate system C1-xy taking the center of the digital photo as an origin; fourthly, using the height line of the building to establish an object space height coordinate axis, and taking an origin of the object space height coordinate axis as a low point of the building and the height direction as the forward direction of the coordinate axis; and fifthly, substituting a coordinate of an image space coordinate axis of a high point of the building into a one-dimensional inverse direct linear transformation formula, and calculating a coordinate of an object space coordinate axis of the high point of the building, wherein the coordinate value is the height of the building.

The invention discloses a scene image splicing method at presence of a moving target. Firstly a scene image of the moving target is acquired by means of a video camera. Then a SIFT algorithm is utilized for extracting an image characteristic. Then characteristic matching between two images is finished by means of the extracted characteristic. Error characteristic matching is eliminated through a RANSAC algorithm, thereby obtaining correct characteristic matching. A geometric conversion model between images is established through a characteristic matching relationship, thereby finishing image registration. Finally fusion processing is performed on the registered image according to an image fusion method. According to the scene image splicing method, based on mobile direct linear transformation estimation and local constant linear fusion method, a splicing ghost caused by the moving target is eliminated through improving image registering precision, thereby improving quality and resolution of a spliced image. The scene image splicing method has advantages of simple and quick algorithm, low workload, high consistency, high efficiency and high registering precision. Furthermore the ghost caused by the moving target is eliminated. The scene image splicing method is particularly suitable for scene image splicing at the presence of the moving target.

The invention provides a digital reconstruction method of a traffic accident scene based on monitoring videos, comprising the following steps of: steps one, continuously extracting frames related to movement positions of a detected vehicle from a monitoring video sequence; step two, calibrating a control point of each frame image in sequence and calculating corresponding values of DLT (direct linear transformation) coefficients; step three, selecting an observation point to acquire image space coordinates of the observation point at different moments and respectively calculating object space coordinates of the observation point at the corresponding moments by combining the values of the DLT coefficients; and step four, carrying out the digital reconstruction of the scene on a movement state of the vehicle to obtain a path line of the vehicle movement, and solving a displacement curve, a speed curve and an accelerating curve of the vehicle movement. The method provided by the inventioncan realize the digital reconstruction on the vehicle movement state in the scene at different road sections, thus the problem of monitoring on the vehicle movement state in a large range is solved, measurement accuracy is improved and cost is reduced.

The invention belongs to the field of visual measurement, and relates to a quick extraction method for the space three-dimensional calibration of a camera. The method comprises the steps: installing a space three-dimensional calibration target before calibration, employing a three-coordinate measuring machine for measurement, and determining the precise three-dimensional position information of the centers of big and small ceramic ball targets; photographing the big and small ceramic ball targets through left and right cameras in a binocular visual system, extracting the feature information of the centers of the big and small ceramic ball targets in images, and solving two-bit pixel coordinates of the centers of the ceramic ball targets; finally solving the internal and external parameters of the left and right cameras on the basis of a direct linear conversion method, and achieving the three-dimensional calibration of the cameras in a three-dimensional space. On the basis of the direct linear conversion method, the method employs the binocular visual measurement system to overcome the limits of a conventional three-dimensional calibration method in the calibration process of the cameras, and achieves the high-precision calibration of the cameras.

The invention provides an imaging model construction method and a calibration method of a linear array rotary scanning camera based on projection transformation, the method comprises that follow steps: based on the imaging principles of linear array rotary scanning camera and frame camera, camera rotation platform parameter, the geometrical relationship between the coordinates of the image pointson the original imaging plane of the linear array rotary scanning camera and the coordinates of the corresponding image points on the virtual frame image with the tangent plane as the imaging plane isdetermined by the positional relationship between the imaging plane of the linear array rotary scanning camera and its tangent plane, so as to determine a new imaging model of the linear array rotaryscanning camera. The calibration method comprises the following steps of: calibrating a linear array rotary scanning camera by using the imaging model, adopting a direct linear transformation methodand a non-linear optimization method according to the obtained control points and corresponding image point coordinates. Compared with the traditional method, the method of the invention is simple andfast, and has higher precision. It is easier to construct nuclear images, so it is more suitable for three-dimensional reconstruction.

Disclosed is an improved identification method for an infrared interactive touch spot. The improved identification method for the infrared interactive touch spot comprises the following steps: (1) coordinate setting is carried out for a camera based on consistency of random sampling; five groups of point correspondence relationships are randomly sampled in a set S, a projection matrix HC corresponding to the five groups of point correspondence relationships is counted out, when a consistent set CC of the HC surpasses a preset threshold value T, then the current point correspondence relationships in the consistent set meet requirements of ideal point correspondence relationships; otherwise, the step is repeated until the consistent set meets the requirements; according to the point correspondence relationships in the consistent set, a final projection matrix H is counted out through direct linear transformation; (2) screen border touch spot detection is carried out through a dynamic parameter method.

The invention discloses an image splicing method and system based on straight line feature and key point feature fusion, and relates to the technical field of image splicing and pattern recognition. In order to extract more features on a low-texture image as much as possible, linear features are introduced, that is, the linear features and key point features on the image are detected at the same time, feature extraction of the to-be-spliced image is enriched, and a moving direct linear transformation algorithm combining the linear features and the key point features is introduced. Linear features and key point features are used for jointly guiding local homography transformation of each grid, in order to prevent the linear features from weakening the alignment capacity of the key point features, the fusion degree of the linear features is balanced through lambda constant coefficients, and it is verified that the moving direct linear transformation algorithm with the balance factor hasbetter low texture image alignment capacity, and the problem of misalignment caused by insufficient SIFT feature points of a low-texture image is effectively solved.

The invention relates to a method for positioning coordinates of magnetic resonance imaging based on direct linear transformation and singular value decomposition. According to the method, a computational process of a matrix of conversion relation between an MRI coordinate system and a coordinate system of an ultrasonic therapy system includes acquiring linear equations of positioning markers through transversal scanning by an ultrasonic probe, and removing the linear equation having the maximal error; acquiring planar equations of the upper and lower surface of the ultrasonic probe through lengthwise scanning by the ultrasonic probe; acquiring coordinates of 10 positioning points; computing a middle conversion matrix H of the MRI coordinate system and the coordinate system of an ultrasonic therapy system through direct linear transformation and singular value decomposition; orderly removing the ith point in the acquired 10 positioning points, and acquiring a corresponding middle conversion matrix Hi using the rest 9 positioning points; and removing two positioning points having the maximal errors, and computing a matrix T of conversion relation between the MRI coordinate system and the coordinate system of an ultrasonic therapy system. Compared with the prior art, the method is high in positioning precision.

The invention belongs to the image processing method field and aims at carrying out different homographic transformation on different planar regions. Therefore, the invention provides an image alignment method based on multi-model geometric fitting and layered homographic transformation. Extracted feature matched points are classified by using a multi-model geometric fitting method; after classification, an image has a base layer and a transition layer, the base layer is processed by using a dynamic direct linear transformation method and the transition layer is processed by using layered homographic estimation to improve the homographic transformation precision. The image alignment method is mainly applied to the image processing occasion.

The fire monitoring technology of the present invention is a fire differential analysis technology, adopts a color CCD camera system, uses the different reflections of the three primary colors of the color image on the early flame, performs differential calculations on the red and blue primary color components, binarizes, and performs differential calculations again, using the flame The growth rate of the area difference image is used to judge the abnormality of the fire, and the direct linear transformation technology is used to locate the fire space, which improves the reliability of fire detection. The existing closed-circuit television system can be used to realize the automatic monitoring, positioning and monitoring of the fire by the computer. Linkage saves.

The invention discloses a large-breadth digital aerial surveying camera image splicing method based on DLT (Direct Linear Transformation) coefficients. The method is mainly used for splicing sub-images of a large-breadth digital aerial surveying camera into a virtual image and comprises the following steps: exposing at the same moment to obtain four sub-images; carrying out connection point matching on the four sub-images in a cross-shaped overlapping region; calculating to obtain an outer orientation element of each sub-image according to connection points among the sub-images and an initial platform checking parameter of each sub-image; determining a relative position relation of four cameras according to the outer orientation elements of the four sub-images which are exposed at the same moment; calculating a DLT platform checking parameter according to the relative position relation; generating a spliced virtual image by the sub-images according to the DLT platform checking parameter. According to the large-breadth digital aerial surveying camera image splicing method based on the DLT coefficients, disclosed by the invention, the complicated data operation processing is avoided, so that the splicing method is simple and is good for the image splicing method to rapidly process.