32results about How to "Restore the truth" patented technology

The invention provides a real-time three-dimensional point cloud reconstruction method based on multiple cameras. The method is characterized in comprising the following steps that: constructing a multi-camera synchronous collection system, carrying out multi-camera system calibration, establishing a projection lookup table and carrying out image collection, object contour segmentation and three-dimensional reconstruction. The invention also provides a real-time three-dimensional point cloud reconstruction system based on multiple cameras. The system comprises a sealed three-dimensional space, a plurality of cameras, a plurality of lamp sources and a computer, wherein the plurality of cameras and the plurality of lamp sources are fixedly installed around the three-dimensional space; the cameras are connected with the computer; and the computer comprises a camera calibration module, a projection lookup table calculation module, an object outline segmentation module and a three-dimensional reconstruction module. The three-dimensional reconstruction disclosed by the invention has the advantages of being higher in accuracy, higher in efficiency, universally applicable and simpler in calculation.

The invention discloses a projector red and blue laser light source system, comprising a first B-LD array, a fluorescence color wheel, a second B-LD array and a B reflector array, and being characterized in that an R area of the fluorescence color wheel is a red light transmission area, a B area is a blue light transmission area, and a G area is coated with green fluorescent powder; the system also comprises an R-LD array opposite to the second B-LD array, and an R reflection DM mirror array arranged in the light ray propagation direction of the R-LD array; a first RB reflector is arranged in the transmission direction of the fluorescence color wheel, a second RB reflector is arranged in the reflection direction of the first RB reflector, an RB reflection DM mirror is arranged in the reflection direction of the second RB reflector, and a light collection rod is arranged in the reflection direction of the RB reflection DM mirror; a G reflection DM mirror is arranged in the reflection direction of the fluorescence color wheel, and the RB reflection DM mirror is arranged in the reflection direction of the G reflection DM mirror. A color shift phenomenon of red laser obtained by the light collection rod is avoided, and the red color of a projector is reproduced very truly.

The invention discloses a switch clock synchronization controller and a control method for a time-triggered network. The switch clock synchronization controller comprises a curing module used for restoring a PCF frame sending sequence, a compression module used for calculating an SM clock deviation average value, an optimal compression point selection module, a local clock module used for correcting a local clock and generating a new PCF frame, a cluster detection module used for detecting a network state and a state machine control module used for coordinating the work of the modules. The curing module imports a transparent clock for identifying the transmission delay and a maximum transmission delay for restoring the sending sequence and moments of the PCF frames for solving the problem of lagged clock synchronization; the compression module adopts a hardware mode to realize a three-level streamline dynamic compression algorithm for improving the calculation precision and efficiency of clock synchronization; and a compression point having the largest member number and closest to a receiving window end moment is used as the optimal compression point to ensure that all nodes in a synchronization state in the network can be detected.

The invention provides a lens for head-mounted display equipment. The lens comprises a double convex lens body and a plano-concave cylindrical lens body, wherein the double convex lens body and the plano-concave cylindrical lens body are arranged side by side; the double convex lens body is arranged on the side close to the human eyes, and the plano-concave cylindrical lens body is arranged on the side away from the human eyes; the double convex lens body comprises an aspheric surface I and an aspheric surface II, and the plano-concave cylindrical lens body comprises an aspheric surface III and as aspheric surface IV; the aspheric surface I is close to the human eyes, the aspheric surface II and the aspheric surface III are oppositely arranged, and the aspheric surface IV is a plane. By the adoption of the head-mounted display equipment, no preprocessing on an original display image is needed, Y-direction compression on the image is performed through the optical lens, and the X-direction size is kept unchangeable. Thus, the aspect ratio of the image is reduced into 4:3 from 8:9 through compression, and the image is displayed on the retina of the human body. In addition, the double convex lens body of the lens is easy to form, the MTF, SPOT and LCA aberration is corrected well, and manufacturing cost and weight of the lens are greatly reduced.

The invention discloses high-adhesion oxidation-resisting UV-curable ink and a preparation method thereof. The UV-curable ink comprises a low polymer, a reactive diluent, a photoinitiator, filler and an additive at a certain weight ratio, wherein the weight ratio of the low polymer to the reactive diluent is 5:3; the photoinitiator is compounded by two kinds of photoinitiators at the weight ratio being 1:1. The invention further discloses a preparation method of the high-adhesion oxidation-resisting UV-curable ink. The preparation method comprises the following steps: at first, weighing all the ingredients in a reactor according to the mass percent, uniformly mixing, placing the mixture into an oven for softening, at last, stirring and dispersing the softened liquid by a high speed disperser, controlling the fineness to 5-20 microns, and canning. The high-adhesion oxidation-resisting UV-curable ink prepared by the preparation method has the advantages that the adhesion force can reach 100%, the curing speed is 8-15 s, the solvent resistance is strong, the deinking phenomenon, caused by that a conventional PE bottle cap is soaked for 4 h in a 1,800 ppm concentrations of peracetic acid solution after being printed with UV-curable ink, and the temperature is controlled to be constant to be 90 DEG C, is overcome, and the sterilization requirements of environmental protection departments are met. The preparation method is simple and feasible, and easy to realize.

The invention relates to a paper and especially relates to a transfer coating ink-jet printing high-gloss photographic paper. The paper comprises a surface coating, a base coating, a raw paper layer and a back coating. The surface coating is an ink absorption water-resistant gloss coating. The base coating is an ink absorption fixation coating. The back coating is a water-resistant coating for balancing two surface tensions of a photographic paper. The base coating comprises light calcium carbonate, a sodium polycarboxylate dispersant, a polyvinyl acetate emulsion adhesive, a polyvinyl alcohol glue solution and a fixer. The surface coating comprises aqueous silica sol and a polyvinyl alcohol glue solution. The back coating comprises polyvinyl alcohol. The photographic paper surface is a high-gloss mirror surface, images obtained by color ink-jet printing have plump colors and realizes real restoration and ink absorption and fixing performances are good.

The invention provides a remote sensing image defogging method based on an edge sharpening cyclic generative adversarial network. According to the method, the foggy remote sensing image is processed by utilizing the deep generative adversarial network, and the ground object information of the foggy remote sensing image can be automatically recovered in a large quantity; secondly, the invention provides a cyclic generative adversarial network model added with an image sharpening mechanism, and the model improves the condition that textures of pictures generated by the generative adversarial network are not clear, and improves the discrimination capability of blurred images; and finally, the VGG16 network pre-training process of calculating the perception consistency loss function by using the model is improved, and the VGG16 network is pre-trained by using the remote sensing image, so that the characteristics of the remote sensing image are better extracted, and the calculation of the perception loss error is more accurate. Compared with an existing remote sensing image defogging method, the remote sensing image defogging method has the advantages of being intelligent, batched and automatic, the quality of the remote sensing image can be remarkably improved, detail information such as remote sensing image textures is protected, and a good recovery effect is achieved.

The invention relates to a prime lens along a direction from an object side to an image side. The prime lens sequentially comprises a first lens (L1) with negative focal power, a second lens (L2) withnegative focal power, a third lens (L3) with positive focal power, a fourth lens (L4) with positive focal power, a fifth lens (L5) with positive focal power, a sixth lens (L6) with negative focal power, a seventh lens (L7) with positive focal power and an eighth lens (L8) with negative focal power. The prime lens has the advantages of being low in distortion, high in pixel, small in size and freeof virtual focus within the temperature range of 0-80 DEG C.

Present invention relates to digit decoding sound track surround sound system earphone. It contains earphone, external power supply plug, and digit decoding power amplifying box accessed between earphone and external power supply plug, said digit decoding power amplifying box power amplifying box having power supply input part, signal input part, decoding part, signal output part and control part, capable of realizing coaxial, optical fiber, and stereo inputting, signal output part containing 5.1 sound track line output, 5.1 sound track power amplification, 5.1 sound track power amplifying outputting, sound track and electronic switch converting. Said invention integrates digital signal decoding and 5.1 sound track power amplifying with sounds accurately positioning, rich gradation, small distortion, and fine sonic balance ability and surround sound effect.

A face handwriting input system based on image acquisition comprises an input panel, an input device, a processing unit and a server platform for storing handwriting input content. Coordinate points are arranged on the input panel to form an input coordinate system, the input device comprises a writing surface, a pressure sensor and an image collector, the writing surface is in contact with the input panel during handwriting operation, and positioning points on the outer edge of the writing surface are contained in the view field of the image collector. The processing unit receives image information of the image collector, calculates handwriting input content and sends the handwriting input content to the server platform. The systrm has the advantages of high handwriting restoration degree, wide writing scene and the like.

The embodiment of the invention discloses a method for creating a three-dimensional dynamic scene. The method comprises the following steps: determining a three-dimensional mapping parameter of at least one image acquisition device by analyzing a historical dynamic image; according to the three-dimensional mapping parameters, mapping two-dimensional attribute information of a first object in a real-time dynamic image into three-dimensional attribute information, the historical dynamic image or the real-time dynamic image comprising a sequence image and / or a video image; adding the three-dimensional attribute information of the first object to a three-dimensional dynamic template; and rendering the three-dimensional dynamic template in real time to obtain a real-time three-dimensional dynamic scene. Compared with a three-dimensional static scene obtained through a traditional three-dimensional reconstruction method, information contained in the dynamic scene is more comprehensive and richer and can serve as a basis for analyzing dynamic object behaviors in the scene.

The invention relates to a road surface depth information model construction method based on vehicle-mounted mobile laser point cloud, belongs to the technical field of road surface disease detection,and adopts a single-lane overlapped section division strategy to control a road surface model. Firstly, adopting an RANSAC algorithm to obtain a consistency estimation plane of single-vehicle road surface point clouds with a certain length, and using the plane as a benchmark to normalize the elevation of original point clouds; smoothing local high-frequency noise and global vibration noise of thepavement by adopting a combined filtering method of median filtering and Butterworth low-pass filtering; and finally, constructing a quadric surface model, taking the filtered and smoothed vehicle-mounted mobile laser pavement point cloud as an observation value, and obtaining an unbiased estimation value of a quadric surface parameter by adopting an iterative least square fitting method for inhibiting low-frequency characteristics. And calculating the distance from the discrete laser points to the curved surface by taking the continuous quadric surface model as a standard road surface to obtain a road surface depth information model. According to the method, the pavement structure can be truly restored, and the pavement depth information model is constructed.

The embodiment of the invention provides a three-dimensional image reproduction method, device and equipment and a storage medium, and the method comprises the steps: scanning a to-be-reproduced three-dimensional image based on a preset viewpoint, and obtaining sky envelope surface coordinates and corresponding sky envelope surface textures of a sky sphere envelope surface relative to the viewpoint in the to-be-reproduced three-dimensional image; Obtaining object image coordinates and corresponding object image textures of the object image relative to the viewpoint in the to-be-reproduced three-dimensional image; And according to the sky envelope surface coordinates, the corresponding sky envelope surface textures, the object image coordinates and the corresponding object image textures, performing restoring to generate a three-dimensional image. Through the technical scheme, the sky envelope surface coordinate and the sky envelope surface texture corresponding to the sky sphere in thestereo image to be reproduced, and the object coordinate and the object texture corresponding to the object image can be obtained, so that the stereo image can be restored more truly and accurately.The problem that the stereo image cannot be recorded, restored and reproduced is solved.

The invention relates to a method for reducing a three-dimensional structure of a non-woven material through a fiber central axis. The method comprises the following steps: extracting a central axis of a fiber through a same-field multi-layer multi-focal-plane non-woven material microscopic image so as to restore the three-dimensional form and position of the central axis in the non-woven material; collecting and fusing multilayer microscopic images; performing the circular segmentation of the fibers, selection of circle center points, calculation of the depth of the circle center points, extraction of the central axes of the fibers and reduction of the closed three-dimensional surfaces of the fibers; extracting the depth of a source point on the central axis of the fiber and the section radius of the fiber at the source point, and then restoring the three-dimensional coordinates of the fiber axis and the closed surface in a spatial interpolation mode. Compared with other reconstruction methods, the method provided by the invention truly restores the relationship between fiber longitudinal directions, and provides a high-quality three-dimensional model for subsequent research on key parameters such as pore size and distribution, pore inner surface area, pore volume, air fluid resistance and pressure difference of the non-woven material.

The invention provides a processing method and processing device of remote sensing image data, which are used for solving problems that it is difficult to distinguish ground objects because real geomorphic characteristics in a shadow region are completely or partially shielded during imaging in the prior art. The method comprises the following steps of: for a preset geographic range, obtaining digital elevation model (DEM) data within a geographic range; obtaining a depth image from remote sensing images within the geographic range by using depth image algorithm and mode identification in the digital image processing technology; and fusing the DEM data and the depth image data by using a DEM data fusion processing technology in a geographic information system. By using the technical scheme provided by the invention, the shadow can be eliminated in a comparatively effective way so that contribution is made to really recovering ground object texture so as to distinguish the ground objects.

The present invention proposes a lens for a head-mounted display device, which includes biconvex lenses and plano-concave cylindrical mirrors arranged side by side; the biconvex lenses are arranged on the side near the human eye, and the plano-concave cylindrical mirror is arranged on the side of the far human eye; The biconvex lens includes aspheric surface I and aspheric surface II, and the plano-concave cylindrical lens includes aspheric surface III and aspheric surface IV; the aspheric surface I is close to the human eye, aspheric surface II and aspheric surface III are opposite, and aspheric surface IV is a plane . With the head-mounted display device designed by the present invention, the original display image does not need any preprocessing, and the image is compressed in the Y direction by the optical lens, and the size in the X direction remains unchanged. As a result, the aspect ratio of the image is compressed from 8:9 to 4:3, which is presented on the human retina. And the lens lens is easy to mold, the MTF and SPOT and LCA aberrations are well corrected, and its manufacturing cost and weight are greatly reduced.

The invention provides a processing method of a stereo image. The processing method comprises the following steps that: pixel data of all compressed pictures of a stereo image are provided and whether a sum of all pixels of each row of each compressed picture forming the stereo image is equal is determined; when the sum of all pixels of each compressed picture is not equal, a boundary of each picture is determined; received pixel data of all compressed pictures are buffered; one compressed pixel datum that has the minimum number of the column is determined as a pixel datum of column n and onecompressed pixel datum that has the maximum number of the column is determined as a pixel datum of column m; if the number of the column of the pixel data among the compressed pixel data is less thanthe number of the column m, the number of the column of pixel data of the picture is complemented as the number of the column m; and then the pixel data of each picture are decompressed; and last pixel data of the column m-column n of the all decompressed pictures are lost. According to the processing method of a stereo image provided in the invention, the boundary can be determined accurately; the image is restored with reality; and the better user experience can be provided. Besides, the invention also provides a stereoscopic display apparatus with utilization of the processing method of the stereo image.

The invention provides a three-dimensional human body size measurement method and system and a computer readable storage medium. The method comprises the following steps: acquiring an original three-dimensional model of a human body to be measured; preprocessing the original three-dimensional model to obtain a preprocessed three-dimensional model; extracting coordinates of skeleton points of the preprocessed three-dimensional model; calculating deformation parameters of a multi-person linear skinned skeleton model SMPL according to the coordinates of the skeleton points and the preprocessed three-dimensional model, and obtaining a standard three-dimensional model of the preprocessed three-dimensional model after SMPL deformation according to the deformation parameters; according to the preprocessed three-dimensional model, non-rigid registration is conducted on the standard three-dimensional model subjected to SMPL deformation, a corrected standard three-dimensional model is obtained, and the corrected standard three-dimensional model is used for three-dimensional human body size measurement. The three-dimensional model is restored more truly, so that a more accurate human body size result is obtained; a measurement mode and a measurement position can be defined in advance, and the method has great flexibility and usability.

The invention discloses an image edge classification detection method based on visual path orientation sensitivity. The present invention utilizes the role of neuron synaptic connections in the centripetal distribution of the optimal orientation of the receptive field, constructs a first-level neuron network sensitive to multiple specific directions, uses image pixels as network input, and records neurons in a certain time window The pulse firing sequence within is calculated, and the discharge frequency is calculated as the network output; the network output in multiple directions is fused and mapped to the gray level to form an edge-sensitive image; for the edge-sensitive image, the lateral inhibition range and inhibition amount in the receptive field are determined , form the second-level neuron network, and output the image after lateral suppression; finally, after threshold processing, the edge detection result is obtained. The invention considers important visual mechanisms such as directional receptive fields and lateral inhibition, simulates the hierarchical processing effect of different hierarchical structures on the visual path in contour perception, and can effectively improve the edge detection performance of low-contrast images.

The present invention proposes a remote sensing image defogging method based on edge sharpening cyclic generation adversarial network. This method uses a deep generative adversarial network to process foggy remote sensing images, and can automatically and massively restore the feature information of foggy remote sensing images. Secondly, the present invention proposes a cyclic generative adversarial network model that adds an image sharpening mechanism. This model It improves the unclear texture of the pictures generated by the generative adversarial network and improves the ability to distinguish blurred images. Finally, the present invention improves the VGG16 network pre-training process for the model to calculate the perceptual consistency loss function, and uses remote sensing images to pre-train the VGG16 network. This allows remote sensing image features to be better extracted and makes the calculation of perceptual loss errors more accurate. Compared with existing remote sensing image dehazing methods, the present invention has the advantages of intelligence, batching and automation, can significantly improve the quality of remote sensing images, protect detailed information such as remote sensing image textures, and achieve good restoration effects.

The invention provides a prediction method and device based on an event network and electronic equipment. The method comprises the steps: obtaining service activity data in any service scene; inputting the service activity data into a service model established based on an event network to trigger the service model to simulate a service scene, wherein the event network is compatible with a discrete mode and a continuous mode, and the service model established based on the event network comprises a logical relationship among service nodes in the service scene and a sequential relationship of event occurrence; according to the process of simulating the service scene according to the logical relationship among the service nodes under the service model and the time sequence relationship of the event occurrence, establishing a touch graph, predicting the event under the service scene, the touch graph being a directed network structure and being used for describing the development direction and the causal relationship of the event. According to the invention, the event network is compatible with the continuous mode, the continuous event can be effectively expressed, the service scene can be truly restored, and the prediction accuracy is improved.

The invention relates to a high-transmittance and low-distortion aspheric zoom lens. A front lens set A, a zooming lens set B and a diaphragm C between the front lens set A and the zooming lens set B are sequentially arranged in an optical system of the lens in the incident light direction; the front lens set A is sequentially provided with a negative crescent lens A-1, a double-concave plastic aspheric lens A-2 and a double-convex lens A-3 in the incident light direction; and the zooming lens set B is sequentially provided with a double-convex plastic aspheric lens B-1, a double-convex lens B-2, a negative crescent lens B-3 and a double-convex plastic aspheric lens B-4 in the incident light direction. According to the lens, chromatic aberration is eliminated by using flint glass with high refraction and low chromatic dispersion and crown glass with low refractive index; by using the plastic aspheric lens, the distortion of the lens is well corrected, and marginal images are reduced truly; and because the lens is made of a little glass, the assembly difficulty of the lens is reduced, and the production cost is improved.

The invention relates to a method for constructing a road surface depth information model based on a vehicle-mounted mobile laser point cloud, which belongs to the technical field of road surface disease detection, and adopts a section division strategy with overlapping single lanes to control the road surface model. First, the RANSAC algorithm is used to obtain the consistent estimation plane of the single-lane road surface point cloud with a certain length, and the height of the original point cloud is normalized based on this plane; then, the combined filtering method of median filtering and Butterworth low-pass filtering is used to smooth Local high-frequency noise and global vibration noise on the road surface; finally, a quadratic surface model is constructed, and the filtered and smoothed vehicle-mounted mobile laser road surface point cloud is used as the observation value, and the iterative least squares fitting method that suppresses low-frequency features is used to obtain the quadratic surface parameters. Unbiased estimation value, using the continuous quadratic surface model as the standard road surface to calculate the distance from the discrete laser point to the surface to obtain the road surface depth information model. The present invention can restore the road surface structure more realistically, and construct the road surface depth information model.

The invention relates to a paper and especially relates to a transfer coating ink-jet printing high-gloss photographic paper. The paper comprises a surface coating, a base coating, a raw paper layer and a back coating. The surface coating is an ink absorption water-resistant gloss coating. The base coating is an ink absorption fixation coating. The back coating is a water-resistant coating for balancing two surface tensions of a photographic paper. The base coating comprises light calcium carbonate, a sodium polycarboxylate dispersant, a polyvinyl acetate emulsion adhesive, a polyvinyl alcohol glue solution and a fixer. The surface coating comprises aqueous silica sol and a polyvinyl alcohol glue solution. The back coating comprises polyvinyl alcohol. The photographic paper surface is a high-gloss mirror surface, images obtained by color ink-jet printing have plump colors and realizes real restoration and ink absorption and fixing performances are good.

The invention discloses a remote controller. A trackball is arranged in a control area of the remote controller. The remote controller comprises a track acquisition module used for acquiring track information of the trackball, a track information processing module used for processing the track information to generate a direction control instruction corresponding to the track information, and an instruction sending module used for sending the direction control instruction. According to the remote controller, the trackball is arranged on the remote controller so that the remote controller not only can conduct direction control in both the horizontal direction and the vertical direction, but also can conduct control in the direction between the horizontal direction and the vertical direction, therefore, direction control in a total space plane range is realized, operation by a user is more sensitive and free, and the operability and controllability of the remote controller are improved.

The invention discloses an image display adjustment method and device, display equipment and a storage medium; the method comprises the steps: obtaining an input original image, and obtaining an RGBA value of each pixel point in the original image; performing parameter adjustment on the RGBA value of the pixel point to generate a new RGBA value, wherein the adjusted parameters comprise one or more of brightness, contrast, saturation and picture color temperature; and generating a new image according to the new RGBA value. According to the invention, the phenomenon that the display effects of the same work are inconsistent due to the difference of display equipment can be eliminated or weakened, and a real image is restored.