87results about How to "Eliminate occlusion" patented technology

The invention discloses a method for segmenting different objects in a three-dimensional scene. The method comprises the following steps of: establishing adjacency relations and a spatial searching mechanism of point cloud data to estimate a normal vector and a residual of each point for the outdoor scene three-dimensional point cloud data acquired by laser scanning; determining the point with the minimum residual as a seed point and performing plane clustering by using a plane consistency restrictive condition and a region growing strategy to form the state that the entire plane building is segmented from other objects in the three-dimensional scene; establishing locally connected search for a plane building region for the segmented entire building part, and clustering the points with connectivity in the same plane by using different seed point rules to realize the detailed segmentation of the plane of the building; and constructing distance label-based initial cluster blocks for the other segmented objects and establishing weighting control restriction for cluster merging to realize the optimal segmentation result of trees. Tests on a plurality of data sets show that the method can be used for effectively segmenting the trees and buildings in the three-dimensional scene.

An image capturing device is equipped with an external parameter estimating unit for estimating external parameters using a distance image obtained by a TOF camera and a luminance image obtained by a CCD camera, a corresponding pixel determining unit for determining a correspondence relationship between pixel positions in the distance image and pixel positions in the luminance image, using previously stored internal parameters of the TOF camera and the CCD camera or the CCD camera, and the external parameters, and an occlusion searching unit for searching for an occlusion region in the distance image, using the correspondence relationship between pixel positions in the distance image and pixel positions in the luminance image.

The invention discloses a broadband wire source for a planar waveguide CTS antenna feed device. The broadband wire source comprises an H-plane sectorial horn antenna, a bias parabolic reflection plane and a flat waveguide. The H-plane sectorial horn antenna and the bias parabolic reflection plane are arranged inside the flat waveguide, and the phase center of the H-plane sectorial horn antenna is arranged at the focus of the bias parabolic reflection plane. According to the invention, cylindrical waves are converted into plane waves, the plane waves are generated by using a reflector antenna principle, and the horn antenna is arranged at the focus of a paraboloid reflector so that a horn antenna radiation field passes through the reflector and generates the plane wave with equal amplitude and the same phase at the caliber surface of the reflector. The working frequency band is wide; the feed source shielding is eliminated and the caliber efficiency is enhanced through a bias; the structure is enclosed so that the edge diffraction effects of a conventional bias paraboloid are removed; and the structure is simple so that adjusting and assembling are easy, late use and maintenance are convenient, and the problem can be solved that the ideal wire source with equal amplitude and the same phase, generated by a conventional waveguide CTS antenna, cannot be realized in an actual project.

The invention discloses a lever lateral mobile damping four-wheeler manned lunar rover folding system and belongs to the field of spaceflight. The system solves the problem that a four-wheeler is easily overturned integrally, difficultly climbs, runs into hollow surfaces, works cumbersomely, is high in energy consumption, low in adaptability and occupies a large space to hinder the implementation of exploration tasks and the like under the complicated geographical conditions on the surface of a star in the prior art. One side of a first rover body (1-1) is welded with two right-angled steel channels (6); the other side of the first rover body (1-1) is in folding connection with one side of a second rover body (1-2) through two middle connection folding mechanisms (8); the other side of the second rover body (1-2) is welding with two right-angled steel channels (6); the first rover body (1-1) is detachably connected with the second rover body (1-2) through a pair of rolling shaft pins (9) and pin slots (15); the lower end face of a bearing plate rover (1) is uniformly provided with four dual-wheel moving mechanisms. The system is used for a manned lunar rover.

The invention particularly discloses an iris image segmentation algorithm based on a nonlinear dimension space. The iris image segmentation algorithm based on the nonlinear dimension space comprises the following steps: inner boundary of an iris is positioned: a dimension evolution characteristic of the nonlinear dimension space is utilized to conduct smooth processing on an original iris image, a hot spot of a pupil area is eliminated, gray level of the pupil area is ensured to be in a floor level of the whole image, pupil detection is achieved through threshold, and positioning of the inner boundary of the iris is achieved; outer boundary of the iris is positioned: the dimension evolution characteristic of the nonlinear dimension space is utilized to conduct smooth processing on the original iris image, sheltering of eyelashes small in geometry dimension on the iris area is eliminated, and accurate positioning of the outer boundary of the iris is achieved. In the process of positioning the outer boundary of the iris, similarly, the dimension evolution characteristic of the nonlinear dimension space is utilized to conduct smooth processing on the original iris image, sheltering of eyelashes small in geometry dimension on the iris area is eliminated, and accurate positioning of the outer boundary of the iris is achieved.

The invention discloses a human face recognition method based on image reconstruction and a Hash algorithm. An image reconstruction model is adopted, a dimension variable ap is used, shields and covers in a human face image are effectively eliminated, and the problems of uneven illumination and image drifting in the human face image are resolved. A Hash matrix is introduced, a high-dimensional data matrix is mapped into a low-dimensional data space, and the calculation amount of the algorithm is largely reduced. Under the condition of guaranteeing recognition accuracy, the problem of multiple inputs in the human face recognition process is effectively resolved, and the recognition efficiency of the algorithm is improved.

The embodiment of the invention provides a method for controlling a cloud mirror, a mobile terminal, front-end equipment and a system, and relates to the field of video monitoring. Monitoring personnel can carry out video monitoring uninterruptedly any time any where, blocking on an image by virtual keys is eliminated, and the monitoring effect is improved. The sliding track of at least one touch point on a touch screen is sampled, and sampling points on the sliding track within sliding time are obtained. A moving instruction is generated according to the position relation between the sampling points and is sent to a network video monitoring platform, the moving instruction is transferred to the front-end equipment through the network video monitoring platform, and the cloud mirror is controlled to move or zoom through the front-end equipment according to the moving instruction. The method for controlling the cloud mirror, the mobile terminal, the front-end equipment and the system are used for improving monitoring portability and improve the monitoring effect.

The invention provides a separated solar observation satellite configuration, which comprises a platform cabin and a load cabin, wherein the platform cabin and the load cabin are connected through a magnetic suspension actuator (1); the separated solar observation satellite configuration also comprises an antenna mechanism; the antenna mechanism comprises an antenna bracket (11); the platform cabin comprises a platform cabin top plate (5); and the antenna bracket (11) is arranged on the platform cabin top plate (5). The separated solar observation satellite configuration can achieve the full physical space isolation between cabins, thoroughly eliminates the micro-vibration effect of the satellite platform on the load, and improves the pointing precision and stability of the satellite by two orders of magnitude. According to the design layout of the antenna bracket in the separated solar observation satellite configuration, the shielding of the field of view of the antenna can be eliminated. The load cabin body in the separated solar observation satellite configuration is attached with a battery sheet, so that a prior fixedly-connected cable is replaced, the quality is reduced, andthe separation of cabin sections is ensured.

The present invention relates, in general, to a microrobot system for intravascular therapy and, more particularly, to a microrobot system for intravascular therapy, which removes thrombus, clots and occlusions that are clogging blood vessels using the fast rotary power of a spherical microrobot having protrusions, thus treating clogged blood vessels.

The microrobot system for intravascular therapy according to the present invention includes a spherical microrobot unit (100′) including a magnet having an arbitrary magnetization direction and having protrusions formed on a surface thereof. An electromagnetic field generation unit (200) sets an alignment direction and a locomotion direction of the microrobot unit to arbitrary directions in the 3D space, generates magnetic fields, and then drives the microrobot unit so that the microrobot unit is rotated by itself or is propelled in the locomotion direction. An imaging unit (300) captures an X-ray image and then tracking a location of the microrobot unit.

The invention discloses a volume rendering method of an automatic identification overlapping structure based on a two-dimensional transfer function. The volume rendering method includes the following steps that firstly, volume data are read in, each voxel is calculated according to two attribute values needed by two-dimensional transfer function space, and the space of the transfer function is constructed; secondly, in order to obtain integrated interesting structures and solve the over-segmentation problem of a watershed algorithm, a transfer function classifier assembly is arranged in the constructed transfer function space and the number of the interesting structures is determined firstly, and then the grey value of the original volume data is automatically corrected; thirdly, for the corrected volume data, a classification result is obtained based on the grey value with the watershed algorithm, and therefore the interesting structures overlapped in the two-dimensional transfer function space can be identified at the same time. The volume rendering method is simple in principle, convenient to understand and capable of being easily achieved in the existing twp-dimensional transfer function space and a visualization system.

The invention relates to a vehicle device for external viewing which comprises a car body, a right rearview mirror and a left rearview mirror, a camera, a display and a control module, at least a backward camera which is needed to be arranged on the external of the car body; at least a display is needed to be arranged on the interior of the car body; the camera and the display are connected with the control module by a lead. When a vehicle is under normal running state, the display, by the control of the control module, can show the images at the right, left and back sides of the vehicle; the invention has the advantages of large viewing angle, no distortion of the images of rear scenes, small blind areas, concentrated observation points, night viewing functions and minimum dependence on external environment.

Provided is a return control system for implementing return to normal status by resolving malfunctioning of a dosing valve or clogging of a liquid transfer line through which urea water flows. A return control system for returning to normal status when a dosing valve (104) malfunctionsis provided with: anomaly detection means for detecting anomalies in the dosing valve (104); and return control means for controlling a supply module (106) in order to return urea water inside the dosing valve (104) to a urea tank (105) when a dosing valve (104) anomaly has been detected by the anomaly detection means.

The invention discloses a gesture recognition method and device based on hyperspectral imaging, wherein the gesture recognition method comprises: acquiring an image of a gesture under common illumination of a visible light source and an infrared light source through a hyperspectral imaging system to obtain a hyperspectral image; The channel hyperspectral image is merged with the target feature toform a single-channel image; the single-channel image is denoised to obtain a denoised image; the denoised image is identified by an image recognition and motion tracking algorithm to obtain a gesturerecognition result. The invention can simultaneously collect the gesture image information of the visible light band and the infrared light band, and effectively combine the target features of the multi-channel hyperspectral image in the gesture recognition, so that the target feature area contrast is enhanced, thereby effectively performing image segmentation and extracting the hand The contourinformation is used for subsequent gesture recognition, which helps to improve the accuracy and effectiveness of gesture recognition.

The invention provides a portable short-focus near-to-eye display system, relates to the technical field of near-to-eye display, and solves the problem of contradiction between optical performance and volume in the existing AR technology. The system comprises a micro display, an inner side lens and a concave surface partial reflector; the lenses close to the eye pupil position are inner side lenses, and the lenses far away from the eye pupil position are concave surface partial reflectors; the micro display is a rotary linear display or a transparent display. Through multiple reflections, the light path is folded, so that the distance between the convex surface part reflector and the concave surface part reflector is shortened, the thickness of the glasses can be reduced to the maximum extent, and lightness and thinness are realized.

The invention discloses a massive curve visualization method for transient analysis of a power system. The method comprises the following steps: firstly, reading time series data involved in transientstability analysis of the power system and performing amplitude alignment on the time series data; converting the time series into a character string by adopting an approximate calculation method, indexing according to a Trie tree, and performing quick clustering by querying the Trie tree; wherein the clustering center is used as time series representativeness and is drawn into a curve; then, calculating the overall density distribution of the time series through a kernel density estimation algorithm by means of the parallel capacity of the GPU; providing interaction tools to classify viewingor select regions of interest to view time series values. According to the method, the density value is mapped to the color space to be visualized through the kernel density estimation method according to the density distribution of the curves in the screen space, and shielding between the curves and visual disorder are eliminated. For thousands of curves, the whole visualization process can be completed within tens of milliseconds, and the fluency of interaction is ensured.

The invention discloses a crystalline silicon solar cell. The crystalline silicon solar cell comprises a crystalline silicon slice substrate. A plurality of transverse PN junctions perpendicular to the end faces of a silicon slice are arranged on the two opposite end faces of the crystalline silicon slice substrate respectively, the PN junctions are composed of P type layers and N type layers respectively, a suede surface and an antireflection layer are sequentially arranged on the front surface of the crystalline silicon slice substrate, and metal electrodes and/or P+ layers are arranged on the back surface of the crystalline silicon slice substrate. A manufacturing method of the crystalline silicon solar cell includes the steps of silicon slice pretreatment, vertical structure treatment, transverse PN junction manufacture, suede surface manufacture, antireflection layer deposition, insulating layer deposition and metallization. According to the crystalline silicon solar cell, incident light loss can be effectively reduced, electrode resistance and contact resistance are reduced, short circuit current, open circuit voltage and fill factors of the solar cell are effectively increased, and conversion efficiency of the solar cell is improved. The crystalline silicon solar cell is high in output voltage and can be used for special occasions.

Provided is a continuous casting method for molten metal, wherein a hollow cylindrical, conical or truncated conical refractory structure provided with one or more side holes in the side wall thereof is disposed above an immersion nozzle in a tundish with the axis of the refractory structure disposed vertically, and molten metal is supplied from the tundish into the immersion nozzle. In said continuous casting method for molten metal, the angle formed by a virtual line extending radially from the center of a circular cross-section in the horizontal direction of the refractory structure and the central axis of the side hole is an angle ([theta]1), the swirling flow of the molten metal supplied into the immersion nozzle is formed by causing the molten metal in the tundish to pass through the side hole, and the flow velocity (Q) of the molten metal, the total opening area (S) of the side hole, the mean internal radius (R) of the circular cross-section in the horizontal direction in a portion in which the side hole is bored, and the angle ([theta]1) satisfy 0.015m2/s<=RQ/SSin[theta]1<=0.100m2/s. By providing a swirling flow imparting mechanism in the tundish, the flow of the molten metal in a mold can be stabilized.

The invention discloses an image segmentation method of a caged laying hen based on an improved active contour model. The S-component images of caged layer images were extracted and analyzed by usingk-Means clustering method to divide the pixels of S-component image into two classes, An S component image is traverse, two class cent functions are obtained, an energy function is set that containsthese two class center functions, the average kernel function and the level set function are introduced into the energy function to form the total energy functional, The standard gradient descent method is used to minimize the total energy functional to obtain the boundary evolution equation. The morphological open operation and Gaussian filtering operation are added to the boundary evolution equation. Finally, the finite difference method is used to iterate the boundary evolution equation until it converges, and the final evolution boundary is the segmentation result of the caged layer image.The invention can quickly and accurately segment the image of the caged laying hen and eliminate the influence of the cage shielding.

The invention discloses a two-dimensional transfer function body data identification method based on a distance transformation watershed algorithm. The method includes: firstly, traversing voxels belonging to a transfer function already set by a user, filling and complementing the voxels, classifying the voxels by utilizing the watershed algorithm according to corresponding negative values of distant transformation, and identifying interested structures overlapped in a transfer function space. A watershed method is over-dependent on computed distance values as a result of over-segmentation of the interested structures, leading to wrong classification or incapability of segmenting structures connected in space, therefore the problem is solved by correcting the distance values according to distance features of the voxels inside the interested structures. On the premise that a new dimension of the transfer function space is not added, each structure is identified by the aid of the distance transformation watershed algorithm, and covering of uninterested structures on the interested structures can be removed. Even when the structures are connected spatially and gray values are similar, satisfactory identification effect can be obtained.

The invention relates to an offset vortex wave generator with a converging function. The offset vortex wave generator comprises a reflective array and an antenna feed source located in a non-oppositearea in front of the reflection array, wherein the antenna feed source is used for radiating spherical electromagnetic waves to the geometric center of the reflective array, and the reflective array is used for converting the spherical waves incident from the antenna feed source into converged vortex waves in a phase compensation mode and transmitting and propagating the converged vortex waves inthe direction perpendicular to the reflective array. The offset vortex wave generator has the advantages that incident spherical electromagnetic waves are subjected to phase compensation through the reflective array, so that the emission of the converged vortex waves with a converging function is realized; the antenna feed source is arranged on the side of the reflective array, so that the shielding of the antenna feed source on the radiating electromagnetic waves is avoided, and meanwhile, the influence of the radiating electromagnetic waves on the antenna feed source is also avoided; and thefocus F formed in front of the reflective array can achieve the effects of converging the generated vortex waves and enhancing the electric field, and the overall planar array structure is simple.