37results about How to "Troubleshoot registration issues" patented technology

The invention discloses a visible light full-color image and multi-spectrum image registering method which comprises the following steps of: performing multi-scale decomposition of the full-color image and multi-spectrum image to generate low-resolution full-color image and multi-spectrum image; extracting and matching SIFT characteristics from the low-resolution images and removing the exterior points; obtaining an initial transformation model by use of the matched SIFT characteristic pair and the iteration re-weighting least square method; performing SIFT characteristic extraction and matching as well as exterior point removal based on image block pairs on the original image by use of the geometric constraint provided by the initial transformation model, and selecting the optimal transformation type from all SIFT characteristic pair sets and obtaining the transformation parameter by use of the iteration re-weighting least square method; and transforming the multi-spectrum image according to the transformation model to obtain the transformed multi-spectrum image.

The invention discloses a vehicle workpiece non-rigid 3D point cloud registration method based on linear mixed deformation. The invention relates to the vehicle workpiece non-rigid 3D point cloud registration method based on linear mixed deformation. The invention aims to solve a disadvantage that the an original ICP method is not applicable when a reference point cloud P is deformed due to gravity of the reference point cloud P or external force for the reference point cloud P which is subjected to non-rigid downsampling at present. The method comprises the steps of (1) obtaining Q and P, (2) constructing a control vector S, (3) constructing a linear mixed deformation model, (4) calculating an initial rigid transformation matrix, (5) constructing a least squares error function, (6) obtaining Delta S, (7) obtaining P', (8) allowing P' to rotate and translate, (9) obtaining the transformation relation between initial source point cloud and the reference point cloud, and (10) judging whether the obtained transformation relation between the initial source point cloud and the reference point cloud satisfies a convergence condition, outputting a result if so, otherwise going to the step (4). The method is used in the field of automobile workpiece registration.

The invention discloses a visible light image and infrared image registering method which comprises the following steps: performing multi-scale decomposition on a visible light image and an infrared image to obtain low-resolution visible light image and infrared image; extracting SIFT (scale invariant feature transform) features, multi-scale corner features and surface point features from the low-resolution image pair; matching the SIFT features and removing exterior points; obtaining an initial transformation model by use of the matched SIFT feature pair, wherein the transformation model refers to any geometric transformation relationship between the infrared image and a high-spectrum image; extracting the multi-scale angular point features and the surface point features based on an image block pair on the original image pair by use of the geometric constraint provided by the initial transformation model; determining a more accurate transformation model according to the initial transformation model and the multi-scale angular point features and surface point features; and transforming the infrared image according to the transformation model to obtain the transformed infrared image.

The invention discloses a point cloud registration method based on a differential evolution algorithm and a TrimmedICP algorithm. The TrimmedICP algorithm can solve the problem of low overlapping rate in point cloud registration, but needs the setting of a proper initial value of a rotation matrix and a proper initial value of a translation matrix at first, otherwise it will fall into the local optimum; the differential evolution algorithm can generate an initial population through a random means and makes the population distributed in all ranges through using mutation, crossover and selection operation; and the differential evolution algorithm and the TrimmedICP algorithm are combined, and therefore, the problem of difficulty in selecting the initial values of the rotation matrix and the translation matrix can be solved, a globe optimal solution can be obtained, and thus, an accurate registration result can be obtained.

A geographic positioning method for ground panoramic images of the present invention first adopts an omni-directional feature quantity for the overall description of the spatial organization orientation relationship, completes rough positioning by retrieving the description vectors at grid points on the satellite image map, and then performs coarse positioning. Positioning determines several matching lines as the control lines when the height information is distorted. Finally, under the constraint that the order of the relative height remains unchanged, the initial value of the iteration is introduced through rough positioning to realize the precise calculation of the outer azimuth elements. Different from the beam adjustment calculation of multiple frames of ground images, this method only needs medium and high-resolution satellite image data and relative elevation information of the scene surface, without any other positioning devices (such as GPS) and without adding artificial control points. Realize the precise external parameter calculation of a single frame of ground panoramic image, which is very suitable for image registration with huge differences in field of view and scale such as sky and earth.

The invention discloses a non-rigid brain image registration method, which comprises the steps of 1, preprocessing a target image according to a reference image to obtain a rough registration image; 2, respectively calculating the candidate characteristic points of the reference image and the rough registration image; 3, constructing characteristic descriptors of the candidate characteristic points, and obtaining matched characteristic points according to the characteristic descriptors; 4, calculating a light stream model of the rough registration image according to the characteristic points to obtain a registration result. Through adopting a descriptor vector based on a Zernike matrix to depict the geometrical characteristics under different scales for describing the regional characteristics of the characteristic points, the characteristic point matching problem during non-rigid registration is solved, and in addition, through fusing characteristic point constraints into the Brox light stream model, the problem of registration of brain magnetic resonance images is solved.

The invention discloses a disposable hard endoscope sleeve lens and a lens alignment method. The lens comprises a rigid head and a flexible sleeve; a connecting end of the flexible sleeve is connected with the rigid head, and the other end of the flexible sleeve is an open free end; a 1:1 nonopaque lens group is arranged in the head to form a sealed front end; the connecting end of the flexible sleeve is provided with a first rigid ring, the free end of the flexible sleeve is provided with a second rigid ring, and first and second limiting devices are arranged on the first and second rigid rings; the flexible sleeve is sheathed on an outer surface of a hard endoscope; and the first and second rigid rings locate the lens group on a position overlapped with the main optical axis of the lens group of the hard endoscope by means of the first and second limiting devices. The method comprises the following steps: keeping the first rigid ring to be coaxial with the lens group through the rigid head; and keeping the first rigid ring to be coaxial with the second rigid ring through the first and second limiting devices. By adoption of the sleeve lens and the alignment method thereof provided by the invention, the sterilization and imaging problems of the existing endoscopes are solved.

The invention discloses a registration method between multi / hyperspectral remote sensing image bands. Through feature extraction and calculation analysis, the gray features of different bands and image space features are combined to solve the registration between multi / hyperspectral image bands. The problem is used for preprocessing of multispectral image color synthesis and hyperspectral image data cube synthesis, thereby improving the universality of multi / hyperspectral remote sensing data.

The invention discloses a visible light image and high-spectrum image registering method which comprises the following steps: performing multi-scale decomposition on a visible light image to form a low-resolution visible light image; generating a high-spectrum image significant waveband image according to the high-spectrum image; extracting SIFT (scale invariant feature transform) features, multi-scale angular point features and surface point features from the low-resolution visible light image and high-spectrum image significant waveband image; matching the SIFT features and removing exterior points; obtaining a transformation model by use of the matched SIFT feature pair; extracting the multi-scale angular point features and surface point features based on an image block pair by taking a registered transformation model of the previous layer as an initial transformation model of the current layer on each layer of visible light image and high-spectrum image significant waveband image; selecting the transformation type and obtaining a transformation parameter according to the initial transformation and the multi-scale angular point features and surface point features in combination with an iteration re-weighted least square method; and transforming the high-spectrum image according to the transformation model to obtain the transformed high-spectrum image.

The invention discloses a geometric constraint-based high-resolution optical satellite staring image registration method. The method comprises the steps of S100, building a free network adjustment directional model of images by utilizing a rational function model of additional image space error compensation; S200, based on the free network adjustment directional model, performing dense matching ona stereoscopic image pair with an optimal rendezvous condition in the images to obtain coordinates of dense homonymy points; S300, generating dense point cloud data, performing gross error point deletion on the dense point cloud data, and generating digital surface model data; and S400, based on the free network adjustment directional model and DSM data, performing geometric correction on the images in sequence. By utilizing geometric locating consistency, high-precision registration of high-resolution optical satellite multi-angle staring sequence images is realized; the multi-angle staringsequence image registration problem caused by satellite platform control inaccuracy and attitude measurement inaccuracy is solved; and a high-precision data basis can be provided for subsequent application.

The invention belongs to the technical field of radars, discloses a multi-station radar signal fusion detection method based on tracking information, and solves the problem in a multi-station radar networking system that the signal fusion detection probability is low. The realization process of the method comprises the steps of: according to a track initial algorithm, obtaining a target initial state estimated value and an original state estimated covariance matrix, and determining a target prediction port door at the kth moment according to the target state estimated value at the (k-1)th moment and the state estimated covariance matrix at the (k-1)th moment; dividing the target prediction port door at the kth moment into a plurality of fusion detection units, calculating the detection threshold of each fusion detection unit in the target prediction port door, and obtaining a measured data set at the kth moment; according to the measured data set at the kth moment, calculating the target state estimated value and the state estimated covariance matrix at the kth moment; and judging whether the track ends. Compared with an existing signal fusion detection method, the target detection probability is improved.