54 results about "Fusion mechanism" patented technology

A system and method for intra-operatively providing a surgeon with visual evaluations of possible surgical outcomes ahead of time, and generating simulated data, includes a medical imaging camera, a registration device for registering data to a physical space, and to the medical imaging camera, and a fusion mechanism for fusing the data and the images to generate simulated data. The simulated data (e.g., such as augmented X-ray images) is natural and easy for a surgeon to interpret. In an exemplary implementation, the system preferably includes a data processor which receives a three-dimensional surgical plan or three-dimensional plan of therapy delivery, one or a plurality of two-dimensional intra-operative images, a three-dimensional model of pre-operative data, registration data, and image calibration data. The data processor produces one or a plurality of simulated post-operative images, by integrating a projection of a three-dimensional model of pre-operative data onto one or a plurality of two-dimensional intra-operative images.

The present invention belongs to the field of biomedicine, and relates to polypeptides capable of inhibiting human coronavirus infections, particularly to polypeptides capable of inhibiting human coronavirus infections in a broad spectrum manner, and applications thereof. According to the present invention, the polypeptides capable of providing broad spectrum inhibition effects for infections caused by more than 2 human coronaviruses are provided based on the conservation property of the S2 region of a coronavirus S protein and the similar fusion mechanism; the test results show that the polypeptides can achieve the commonality of the human coronavirus, ie., the similar HR region and the same fusion mechanism mediated by the coronavirus S protein so as to provide a series of HCoV-EK polypeptides, wherein the polypeptides provide good inhibition effects for currently popular human coronaviruses, and further provide good inhibition activity for SARS virus (RsSHC014-CoV or RsW1V1-CoV) possibly infecting human; and the polypeptides of the present invention can provide the prevention and treatment candidate drug for the currently popular human coronaviruses and the novel human coronavirus possibly emerging in the future.

The invention discloses a video semantic representation method and system based on a multi-mode fusion mechanism and a medium. Feature extraction: extracting visual features, voice features, motion features, text features and domain features of a video; Feature fusion: performing feature fusion on the extracted visual, voice, motion and text features and domain features through the constructed multi-level hidden Dirichlet distribution topic model; And feature mapping: mapping the fused features to a high-level semantic space to obtain a fused feature representation sequence. The model utilizesthe unique advantages of the theme model in the semantic analysis field, and the video representation mode obtained through model training on the basis of the model has ideal distinction in the semantic space.

The invention provides a vehicle category identification method based on multi-channel video information fusion of license plate information and vehicle profiles. The method comprises the following steps of 1), obtaining video images by adopting a plurality of cameras mounted in different positions, and obtaining a vehicle side image and a vehicle head or tail image; 2), detecting, tracking and separating the vehicle images; 3), generating classification results by adopting corresponding methods for the images acquired by the cameras mounted in the different positions; and 4), comparing detection results of different methods, and adopting a simple voting information fusion mechanism. According to the vehicle category identification method based on the multi-channel video information fusion of the license plate information and the vehicle profiles, provided by the invention, the accuracy is effectively improved.

The invention discloses a multi-strategy adaptive lane line detection method based on stereoscopic vision, and the method is based on the stereoscopic vision, and improves the robustness of a lane line detection task through the fusion of the three-dimensional depth information of a disparity map and the RGB semantic information of a two-dimensional image. In the detection stage, a road surface area is adaptively segmented by priori knowledge of three-dimensional environment information to serve as a lane line estimation area, obstacles which are not related to detection are filtered out, andthe problem of interference is solved. In the lane line extraction stage, RGB semantic information of a two-dimensional image is used, and a multi-strategy fusion mechanism and a sliding window technology in an image time domain are adopted to further eliminate interference of areas such as tree shadows, road signs and local fouling related to detection. A three-order Bezier curve model is adoptedin the lane line fitting process, and order characteristics of different lane lines can be better expressed. According to the method, the robustness of a vision-based lane line detection task is greatly improved, and the data point search speed and accuracy are improved.

The invention belongs to the technical field of computer image analysis, and particularly relates to a video behavior identification method based on a time sequence causal convolutional network. The method comprises the following steps: extracting spatial-temporal semantic feature representation from a plurality of video clips by using a sequential causal three-dimensional convolutional neural network to obtain a predicted behavior category; and performing modeling on the frame sequence up to the current moment to extract space-time high-level semantic features for behavior positioning and precision prediction, wherein a fusion mechanism of spatial convolution and time sequence convolution and a causal space-time attention mechanism are designed. The method has the advantages of high precision, high calculation efficiency, real-time performance and the like, is suitable for online real-time video behavior detection and analysis tasks, and can also be used for offline video behavior recognition, abnormal event monitoring and other tasks.

The invention discloses a wind turbine generator bearing fault diagnosing method based on a deep joint adaptation network. The wind turbine generator bearing fault diagnosing method based on the deepjoint adaptation network comprises the following steps: 1), establishing a multi-element fusion database; 2), establishing a deep joint adaptation model; 3), establishing a wind turbine generator bearing fault diagnosing model of the deep joint adaptation network; 4), establishing a multi-GPU cluster computing system. With the wind turbine generator bearing fault diagnosing method, according to distribution difference characteristics between training data and target data during monitoring under different actual working conditions, an inter-domain invariant characteristic representing and probability distribution difference correcting mechanism is explored, a fault target recognition strategy based on an inter-domain joint distribution adaptation and common characteristic deep learning fusion mechanism is provided, advantages of a deep self-encoding network can be utilized, the characteristics are not required to be artificially selected, better, abstract and advanced characteristics can be automatically extracted, and the computational complexity of a classification algorithm is reduced; the wind turbine generator bearing fault diagnosing method is especially suitable for a multi-noise, large-data, complex-structure and dynamic system.

The invention discloses a design method of a multi-dimension attribute data oriented multi-layered clustering fusion mechanism. The method comprises the following steps: 1) converting a data set into a matrix form, and preprocessing data; 2) according to data index attribute characteristics, extracting an optimal reference standard, and carrying out normalization processing on the data; 3) calculating a grey correlation degree, generating a similar matrix of the grey correlation degree, and then, carrying out grey correlation degree clustering to obtain a primary clustering result; 4) according to the primary clustering result in the step 3), adopting a rough set theory to establish a decision table system; 5) calculating an attribute significance information entropy of the decision system for each clustering member; 6) setting a weight for each clustering member; and 7) according to the calculated weight, adopting a probability method to calculate a probability of each data object in each class level to which the data object belongs, selecting the class level where the data object belongs to when the probability is highest to serve as the class level to which the data object belongs to, and obtaining a final clustering fusion result.

The invention relates to an information-fusion-mechanism-based multi-robot dangerous odor source positioning method. In the conventional methods, information is incompletely utilized, and communication constraints are not taken into account, so a multi-robot system may be locally convergent. The method provided by the invention comprises the following steps of: first establishing an information fusion matrix and an odor source position mean value distribution probability model, extracting mean value information, and storing the extracted mean value information into the information fusion matrix; then establishing odor source position probability distribution according to the mean value information, and extracting position information; and finally controlling robots to move to a target position by adopting spiral paths as motion trails of the robots based on probable positions and current position of an odor source, and continuously updating position information, corresponding to maximum concentration, obtained by the robot group and the position information, corresponding to the maximum concentration, obtained by individual robots until a given position is obtained. By the method,the multi-robot system can rapidly position the dangerous odor source, and effectively provide the probable positions of the dangerous odor source under the conditions of communication constraints.

The invention relates to an LTE-U idle channel evaluation method based on a multi-slot fusion mechanism, and belongs to the technical field of LTE-U (LTE in unlicensed spectrum) in mobile communication. An LTE-U system works in an unauthorized frequency range, and before the system access a channel, idle channel evaluation is needed generally. The LTE-U system divides the whole CCA window into multiple slots; sampling data of the multiples slots is used to generate a sampling matrix, a covariance matrix of the sampling data is calculated to obtain characteristic values, the characteristic values in ascending order are used to carry out unequal weighting on the corresponding slots, and the channel state evaluation performance when the CCA window is not completely aligned with a data packet is improved. The idle channel evaluation method can be used to improve the accuracy of channel state determination effectively, reduce the possibility of data packet collision among different systems, and the improves the spectral efficiency that the LTE-U system and a WIFI system coexist in the same channel in the unauthorized frequency range.

The invention discloses a model for predicting a disease risk by applying a convolutional neural network. The model is a four-layer convolutional neural network model which comprises a first layer, asecond layer, a third layer and a fourth layer; the first layer is composed of event matrixes built by EHRS, the second layer is a single side convolutional layer for extracting phenotypes the EHR matrixes in the first layer, the third layer is the maximum pooling layer with important phenotypes in the phenotypes extracted in the second layer, and the fourth layer is a softmax predicted whole connection layer. In the event matrix in the first layer, the horizontal dimension corresponds to a time stamp, and the vertical dimension corresponds to an event value; for combining patient EHR time smoothness, three different time fusion mechanisms are researched in the model; early fusion, late fusion and slow fusion are performed, and an advanced CNN framework with time fusion is built.

The invention discloses a human activity change clue monitoring method oriented to protected area supervision. The method comprises the steps of data preparation and registration, human activity automatic change detection, pseudo change pattern spot removal, change result interactive collection and editing and human activity change detection result display. The invention provides a deep fusion mechanism based on a remote sensing and GIS technology, and a whole set of remote sensing change detection technology system combining automatic detection and interactive collection and editing is formed by constructing change detection capabilities such as big data analysis, image and GIS analysis and multi-temporal image comparison and analysis. The efficiency of change detection service is further improved, human activities in the protected area can be rapidly and comprehensively detected, a basis is provided for timely discovering whether human activity changes affect the protected area or not, and technical support is provided for strictly protecting the protected area.

The invention provides a multi-label long text classification method introducing a multi-path selection fusion mechanism, and relates to the technical field of multi-label long text classification based on a sequence-to-sequence architecture. According to the invention, the effect of completing multi-label long text classification based on a sequence-to-sequence architecture is improved; data published by challenging contest is learned based on a certain machine; the title data and the description data are spliced to obtain long text data; for undescribed data, a problem is copied as a description, then low-frequency word removal preprocessing is carried out on the data to obtain more effective data, a converter model added with a multi-path selection fusion mechanism is adopted for the obtained data to generate a label sequence for an input long text, and redundant information is effectively removed during decoding. Under test data, the recall rate of the label sequence generated by the model is 0.5% of the recall rate of a model which is not added with multi-path selection fusion; and the accuracy and the F1 value are improved by 1%.

The invention provides a deep learning super-resolution method based on an enhanced upsampling and discrimination fusion mechanism, which comprises the following steps of: directly extracting original features of a low-resolution input image by using a single layer based on a deep learning convolutional layer in a residual error branch; using 6 cascaded multi-layer feature fusion-based circular convolution units to extract deep features, and enabling the deep features output by each circular convolution unit to be subjected to up-sampling through 6 deconvolution layers; fusing the up-sampled high-resolution features by applying a feature fusion mode with a discrimination mechanism, and performing dimensionality reduction on the up-sampled features by using a single-layer convolutional layer to obtain a residual error of the high-resolution image; carrying out up-sampling on the low-resolution image by using a bicubic interpolation method in a mapping branch; and adding the mapping and the residual of the high-resolution image pixel by pixel to obtain a final high-resolution image. According to the invention, an up-sampling opportunity is provided for low-resolution characteristics of each stage, and higher reconstruction accuracy is obtained.

The invention discloses a support-free 3D printing machine. The support-free 3D printing machine comprises a supporting shell, wherein the supporting shell is connected to a power output component of an X-axis translating mechanism; a fusion mechanism which is used for fusing consumable items which enter the supporting shell, a multi-sprayer assembly and a power supply mechanism are further mounted in the supporting shell; the multi-sprayer assembly is used for spraying out the fused consumable items and has a function of switching over multiple sprayers; the power supply mechanism provides power for switching of the sprayers of the multi-sprayer assembly; connecting columns which are uniformly distributed in the peripheral direction at intervals and are connected with the sprayers are arranged on a rotary seat; a plane on which the central axes of the connecting columns are located and a plane on which the central axis of the rotary seat is located are base surfaces; an included angle between each two adjacent base surfaces is the same and is delta; a power output component of the power supply mechanism is connected with the rotary seat; the rotation angle of the driving rotary seat is an integral multiple of delta; and printing on an object in the horizontal direction can be realized, so that support-free printing of a suspension structure is realized.

The invention discloses a welding mechanism of a position punch fixture glue piece. The welding mechanism comprises the glue piece and an adhesive tape fusion mechanism and an impact mechanism, wherein a low-power speed regulated motor drive and a high-precision splitter drive are utilized in the glue piece and the adhesive tape fusion mechanism, a trimming die and a bottom die can carry out 360-degree peripheral and 90-degree slitting rotation operation; the impact mechanism is internally provided with a high-precision linear slide rail and a linear sliding block so as to enable the impact mechanism to achieve the switchover of a die cutting region and a heat seal region, and a cylinder is driven by die cutting so as to drive whole upper structure to ascend and descend; and when an ironware stamping die is in operate, the glue piece and the adhesive tape fusion mechanism bear the impaction of the impact mechanism, thereby finishing the precise punching process of various elastic and non-elastic adhesive tapes. The welding mechanism of the position punch fixture glue piece provided by the invention is used for fusing a fixture plastic glue piece and the adhesive table after elastic adhesive tape position punch is carried out; and the welding mechanism is especially suitable for product production processes such as punching and fusion of a TPU (Thermal Receiver Unit), punching and fusion of PVC (polyvinyl chloride) flexible glue, punching and fusion of EVA (ethylene-vinyl acetate) and the like.

The invention relates to a network packet loss rate boundary characterization method under a multiple-data stream fusion mechanism. The method is characterized by comprising the steps of designing an arrival model, a service model and a corresponding boundary function on the basis of the distribution characteristics of flow backlog, modeling a coding node based on a fusion service mechanism, deriving the boundary function and the arrival curve of the node in a phased manner, and establishing the corresponding relationship between the system packet loss rate boundary performance on one hand and the system cache capacity, the data flow arriving the node, the service ability and the quality of service demand on the other hand based on the algebra theory under both conditions of full-fusion opportunity and no-fusion opportunity, thus getting the packet loss rate boundary performance under multiple-data stream fusion service. According to the invention, the arrival process and the service process are modeled based on a cumulative flow center arrival curve and a strict random service curve for the first time, and the behaviors and characteristics to which the two different random processes pay close attention are embodied. The method is more practical, and meets the needs of operating units for network configuration.

The present invention provides a fusion mechanism of wireless self-organizing network operating different routing protocols. The aim of the mechanism is self-adaptingly execute network fusing through perception in the gradual closing process of two networks. The node finds the other network through real-time interception. The network authentication is executed through handshaking information. A uniform routing protocol is confirmed through a priority comparing algorithm and the conflicted IP is discovered through Hash algorithm. The mechanism simply and effectively realizes the fusion of two networks and finally unifies routing protocols. The IP conflict is settled and the long-time communication of two networks is facilitated.

The invention discloses a full-automatic production system for paper containers. The full-automatic production system comprises a front rack and a rear rack. A paper suction mechanism, a left splash-proof sheet cutting and fixing mechanism, a right splash-proof sheet cutting and fixing mechanism, a heating mechanism and a page holding butt fusion mechanism are arranged on the front rack in sequence, and a multi-station indexing rotary tray is arranged on the rear rack and provided with a top cup cylinder station, a cup bottom paper punching station, a primary hot air station, a secondary hot air station, a rotary bottom winding station, a rotary bottom rolling station and a cup falling station in sequence. The full-automatic production system of paper containers automatically completes the processes of paper feeding, left and right splash-proof sheet fixing, cup barrel butt fusion, bottom paper cutting, bottom rolling, finished product collection and the like, and professional equipment is provided for large-scale production of butterfly cup paper containers.

The embodiment of the invention provides a message processing method and device, electronic equipment and a computer readable storage medium, and the method comprises the steps: obtaining an interfacemessage; and according to the multi-layer control tree, routing the interface message to an interface processing logic corresponding to the tail end interface or an interface processing logic corresponding to a single control in any layer of interface layer by layer. According to the embodiment of the invention, a loose fusion mechanism among multiple layers of interfaces is realized, so that theinterface engine of each layer of interface is conveniently and independently developed, and the flexible customization of the embedded interface is realized. In the whole message transmission and processing process, the problem of coordination and synchronization among multiple threads is not involved, and the problems of time consumption or jamming and the like caused by coordination and synchronization of the multiple threads are avoided.

The invention provides a second-hand vehicle pricing method, a device, equipment and a computer readable medium. The method comprises the following steps: a parameter acquisition step, namely obtaining vehicle parameters of a second-hand vehicle to be priced; a data processing step, acquiring reference vehicle parameters based on the vehicle parameters; a pricing model generation step of constructing n vehicle pricing models based on the vehicle parameters and the reference vehicle parameters, and weighting the n vehicle pricing models by using BOOST to obtain a vehicle weighted pricing model,and a pricing step of obtaining pricing of a second-hand vehicle by using the vehicle weighted pricing model based on the vehicle parameters. According to the invention, a data set expansion mode isprovided; according to the method, the pricing of the second-hand vehicle is more objective and accurate, a single-model BOOST fusion mechanism is provided, the pricing result of the fused weighted pricing model is more accurate, a profit rate calculation mode is provided, the retail price and auction price of the second-hand vehicle can be estimated at the same time, and the pricing efficiency and accuracy are improved.

The invention relates to the technical field of UAVs, and discloses a UAV based image splicing method and device. The method comprises: acquiring an image to be spliced, wherein the image to be spliced comprises a first image and a second image; extracting feature points of the image to be spliced; performing rough matching on feature point pairs of the first image and the second image; using theedge detection algorithm to perform edge detection; removing mismatched feature point pairs; and adopting a fade-in and fade-out fusion mechanism to splice the first image and the second image. The extracted image features are matched, a transformation matrix between the images is estimated, and then the transformation matrix is used to align the images, the "ghosting" phenomenon of the images iseffectively eliminated, the matching speed and splicing speed are improved, the splicing accuracy is improved, and hardware requirements are low.