1259 results about "Research Object" patented technology

The invention relates to the object verification technology in the machine vision field, and especially relates to a multi-scale small object detection method based on deep-learning hierarchical feature fusion; for solving the defects that the existing object detection is low in detection precision under real scene, constrained by scale size and different for small object detection, the invention puts forward a multi-scale small object detection method based on deep-learning hierarchical feature fusion. The detection method comprises the following steps: taking an image under the real scene as a research object, extracting the feature of the input image by constructing the convolution neural network, producing less candidate regions by using a candidate region generation network, and then mapping candidate region to a feature image generated by the convolution neural network to obtain the feature of each candidate region, obtaining the feature with fixed size and fixed dimension after passing a pooling layer to input to the full-connecting layer, wherein two branches behind the full-connecting layer respectively output the recognition type and the returned position. The method disclosed by the invention is suitable for the object verification in the machine vision field.

The invention discloses a region-level and pixel-level fusion saliency detection method based on convolutional neural networks (CNN). The research object of the method is a static image of which the content can be arbitrary, and the research goal of the method is to find a target striking the eyes of a person from images and assign different saliency values to the target. An adaptive region generation technology is mainly proposed, and two CNN structures are designed and used for pixel-level saliency prediction and saliency fusion respectively. The two CNN models are used for training a network model with the image as input and the real result of the image as a supervisory signal and finally outputting a saliency map of the same size as the input image. By means of the method, region-level saliency estimation and pixel-level saliency prediction can be effectively carried out to obtain two saliency maps, and finally the two saliency maps and the original image are fused through the CNNs for saliency fusion to obtain the final saliency map.

A depth significance-based remote sensing image rapid retrieval method is disclosed and belongs to the field of computer vision. The method disclosed in the invention specifically relates to technologies such as in-depth learning, significance object detection, image retrieval and the like. According the method, remote sensing images are research objects, and in-depth learning technologies are used for researching a remote sensing image rapid retrieval method. A full convolution neural network is adopted for constructing a multitask significance object detection model which is used for doing significance detection tasks and semantic segmentation tasks at the same time, and depth significance characteristics of the remote sensing images are learnt in network pre-training processes. A depth network structure is improved, a Hash layer fine tuning network is added, and binary system Hash codes of the remote sensing images can be obtained via learning. Significance characteristics and the Hash codes are used comprehensively for similarity measurement. The method disclosed in the invention is of high application value for realizing accurate, highly efficient and feasible retrieval of the remote sensing images.

The invention discloses a motion recognition method based on a three-dimensional convolution depth neural network and depth video. In the invention, depth video is used as the object of study, a three-dimensional convolution depth neural network is constructed to automatically learn temporal and spatial characteristics of human body behaviors, and a Softmax classifier is used for the classification and recognition of human body behaviors. The proposed method by the invention can effectively extract the potential characteristics of human body behaviors, and can not only obtain good recognition results on an MSR-Action3D data set, but also obtain good recognition results on a UTKinect-Action3D data set.

The invention provides a portable ship water gauge automatic detection and identification method based on machine vision. Video data of a ship water gauge in the overwater environment are used as a main research object, the machine vision and the image processing theory are used as main means, so as to detect and identify the scale of the ship water gauge and perform statistical analysis on detection results of continuous video frames, and finally, a ship draft value is obtained. According to the system, the relatively objective ship draft value can be obtained in accordance with image data, the detection results are visually displayed, the workload of ship draft detection personnel can be reduced, the measurement stability and accuracy of the ship water gauge are increased, and an objective measuring means and reference data are provided for determining ship draft.

An multi-modality imaging system for imaging of an object under study that includes a magnetic resonance imaging (MRI) apparatus and an MRI-compatible single-photon nuclear imaging apparatus imbedded within the RF coil of the MRI system such that sequential or simultaneous imaging can be done with the two modalities using the same support bed of the object under study during the imaging session.

The invention discloses a multilayer quotation recommendation method based on a literature content mapping knowledge domain, and belongs to the field of information recommendation and intelligent information processing. The method comprises the following steps: firstly, obtaining the query requirement of a user, wherein the query requirement consists of the key words of the title and the digest of a thesis which needs to recommend a quotation thesis or quotation literature; then, on the basis of the literature content mapping knowledge domain, expanding and querying a retrieval word, wherein the mapping knowledge domain consists of the research object word and the research behavior word node of the literature, and edges which express various semantic relations including synonymy, synonym, an up and down position, part-whole, juxtaposition and the like; and finally, constructing the inverted index of the literature in a data set, selecting a candidate quotation, calculating the similarity between the candidate quotation and query, and adopting a gradient progressive regression tree to carry out quotation recommendation. The method carries out multilayer quotation recommendation on the basis of the literature content mapping knowledge domain, enlarges the range of the candidate quotation, accurately expresses the research object and contents of the thesis, improves efficiency for users to obtain a relevant literature and has a wide application prospect.

The invention discloses a method for the calculating ice coating thickness of a transmission line by utilizing video image processing technique, belonging to the technical field of digital video image processing or online monitoring of the transmission line. The method takes digital image intercepted from a video flowing of the transmission line which is transmitted into a surveillance center as the object of study and processes the image by methods of gradation of image, two-dimension image segmentation, filtration, regional mark and the like in advance. In the process of pretreatment, the image is segmented by adopting a new two-dimension varimax based on simulated annealing genetic algorithm, and the image of the transmission line is marked by adopting eight connected region marking method. Finally, by the contrast and calculation of the pixels of the images which are obtained before and after the ice coating of all the transmission leads, an average value is obtained, and the ice coating thickness is further calculated. When the ice coating thickness of any of the transmission leads exceeds the prescriptive safety range, alarm is given, so that deicing measure is adopted in time, thus providing security for the safe running of an electric power system.

The invention relates to a method for modifying aramid fiber by using carbon nano tubes, which comprises the following steps of: selecting an epoxy resin/ aramid fiber composite material as a research object, selecting the carbon nano tubes as a surface modification material, forming free amino groups in a certain ratio on the aramid fiber by a chemical modification method, soaking the aramid fiber into N-methylpyrrolidone solution (NMP), and reacting the solution and the surface carboxylation carbon nano tubes with the aid of ultrasonic, wherein part of carbon nano tubes enter the interior of Kevlar fiber, and the other part of carbon nano tubes are fixed on the surface of the fiber through amido bonds formed by the carbon nano tubes and amino, thus, a aramid fiber complex inside and outside which the carbon nano tubes are doped is formed. Single fiber tensile test results and single fiber pull-out test results show that the mechanical property of the modified aramid fiber and the interface bonding strength with the epoxy resin are improved.

The invention relates to the technical field of logistics, particularly to a logistics center storage position allocation method. The method comprises the following optimization steps: calculating goods correlation; clustering and analyzing goods; allocating clustered goods to storage areas; allocating goods to specific goods positions and finely adjusting the storage of special goods according to the characteristics of the goods. The method reduces the operating time for a warehouse and improves the operating efficiency of the shelf-type warehouse without increasing equipment investment, by optimizing a logistics center storage position allocation management system, using the shelf-type warehouse center as a research object, optimizing the goods position allocation of the warehouse and beginning with optimization management and dispatching of the warehouse.

The invention provides a digital ship model plane motion organization experimental method based on computation fluid dynamics CFD software FLUENT. The method includes applying a FLUENT pre-process software GAMBIT to establish a research object model and a control domain; arranging a triangle grid on the model surface, then arranging a unstructured grid in the control domain; setting the boundary condition, joining an user-defined function UDF document, introducing dynamic grid technology, using a finite volume method based on the complete unstructured grid, realizing a pure swaying movement, a pure rising and falling movement, a pure oscillatory movement, a pure pitching movement and a pure rolling movement carried on by a plane motion organization experiment; applying a science computation software MATLAB to do Fourier expansion of a force and a moment coefficient obtained by the FLUENT, being fitted by EXCEL least squares method, obtaining a hydrodynamic force coefficient of a vertical plane and a horizontal plane as well as the related hydrodynamics analysis by dimensionless. The invention carries on the digit plane motion organization experiment using the CFD software, can satisfy the request of hydrodynamic force data in the design initial period of a submersible.

The invention discloses a spatial digital plotting and 3D visualization method for a private garden. The method comprises the following steps that S1) data collection is carried out by obtaining holographic data of a plotting research object as well as holographic data of a control target via a 3D laser scanner, and close-range multi-visual-angle shooting measurement is carried out; S3) a point cloud model is established based on 3D laser scanning, concretely, after de-noising, coordinate matching, sampling rarefying and redundancy reduction, a spatial point cloud model of the plotting research object is established based on 3D laser scanning, S3) a 3D model is established based on a close-range multi-visual-angle image by obtaining the point-cloud data model according to the real-scene colors after splicing; S4) multi-source 3D model data fusion and model generation are realized; and S5) a 2D linear drawing is generated on the basis of the 3D laser scanning holographic data.

In a system for providing an illegal use research service for image data, the image data received from a terminal apparatus is registered as a research subject, a research condition for research whether or not the image data being registered is illegally used by a Web site on the network is set, image data that are identical or similar to the image data being the research subject from a web site on the network are searched for based on a feature amount and the research condition, and a search result is informed as a research report to the research client using said terminal apparatus.

The invention discloses a remote sensing image classification method based on image block active learning and belongs to the technical field of image information processing. The method comprises the following steps of remote sensing image blocking, initial sample selection, classifier model training, active learning sample selection, training sample set and classifier model updating, classification process iteration; image block classification prediction, and conversion of a block classification result into a pixel classification result. The remote sensing image classification method serves an image block as a research object, compared with a traditional remote sensing image classification method based on active learning of pixel points, under the same experiment condition, the classification result of the image is more accurate, a block sample screened out by the active learning can more rapidly and accurately conduct manual annotating, constitutive properties of the classified image are stronger, spots directly brought by classification of the pixel points are greatly reduced, and the better visual effect is brought to people.

The invention discloses a microblog water army identifying method based on a probabilistic graphical model. The microblog water arm identifying method comprises three steps as follows: quantizing user attributes and conventional microblog using behaviors into characteristic indexes which are respectively attribute characteristics and behavior characteristics of the user; regarding probability that the user belongs to the water army as a hidden variable of a graph by constructing a probability graph structure between the attribute characteristics and the behavior characteristics; and predicating the probability that other users belong to the water army by utilizing parameters of a sample learning graph model as well as the learned probability graph model. The invention provides the microblog water arm identifying method which is automatic, effective and quick by searching how to automatically identity water army users in the microblog platform, so that the identification is effectively carried out under a large data volume condition of the microblog. Moreover, the microblog water army identifying method has a good practical value and an extensive application prospect in the technical field of data mining.

The invention discloses a determination method for a partition D-STATCOM integrated optimization configuration scheme of a large-scale power grid. The method comprises the steps of first starting from a dynamic analysis method theory of a non-analytic complex power system, establishing a power system non-linear equivalent model, rapidly and accurately calculating system integration dynamic equivalence impedance and load static equivalence impedance of an a load node of a study area to obtain limit margin; enabling areas with total loads ranking at the top to be weak areas according to researched network load flow calculation analysis results, enabling dispatching partitions in the weak areas to be study objects, comparing voltage stability strength indexes of every partition to determine weak partitions so as to enable the weak partitions to serve as configuration study objects, utilizing the fact that an average value of substation impedance modulus margins of the partitions is a threshold value to determine reactive power compensation to-be-selected point sets, shrinking configuration point hunting zones, and comprehensively utilizing trace sensitivity index maximum points to determine configuration points; and finally utilizing an optimal model considering voltage dip risk cost to determine an optimal capacity.

The invention relates to an active distribution network energy-storage system dynamic planning method which studies active distribution networks which contain distributive photovoltaic and energy-storage systems. The method herein optimizes an output strategy of the energy-storage system on the basis of a bilevel programming model in combination with the shortest path theory so as to determine the installed capacity of the distributive photovoltaic and energy-storage system under the optimal output strategy of the energy-storage system. For the bilevel programming model, upper level programming determines the installed capacity of the distributive photovoltaic and energy-storage system with the minimum annual general fee of a power distribution network as the goal and generates a planning scheme, and the lower level planning optimizes the output strategy of the energy-storage system in one operation cycle and in one day with the minimum typical daily operation fee and load peak-valley difference of the power distribution network as the goal. Compared with prior art, according to the invention, the method herein can effectively improve operation characteristic of the active distribution network and optimization configuration of the distributive power supply and energy-storage system.

A mass point cloud spatial management method based on octree-like encoding takes the massive point cloud data as the research object, and is a hybrid point cloud index structure which combines the spatial segmentation of KD-tree to achieve efficient management of mass point cloud. The method is characterized in that firstly, the space of point cloud is partitioned by KD tree, and the subspace is further partitioned by octree, and the octree-like index structure is established; the traditional linear octree coding is improved and the optimized coding method is used to encode the space in orderto achieve better spatial management and neighborhood search. The experimental results and comparative analysis show that octree-like structure can make the overall structure of data organization morereasonable, improve the access efficiency and reduce the occupancy of memory space. The method not only improves the speed of traditional KD tree index, but also improves the problem that traditionaloctree occupies too much space and takes too long to search neighborhood, and realizes the reasonable management of massive point cloud space.

The invention discloses a man-machine cooperation path planning method for an agriculture mobile robot based on a cloud model. The method comprises the following steps: dynamically guiding an algorithm A*(DGA*) to perform man-machine cooperation path planning, combining professional knowledge, hobbies and the like of people to form an expert system, introducing online uncertainty reasoning based on the cloud model, and realizing a rapid path planning strategy. The invention achieves the following beneficial effects that the agriculture mobile robot is taken as a research object, a large amount of experimental data are processed at the cloud by using the online cloud reasoning method based on man-machine cooperation, the time delay is greatly reduced, and the real-time reliability of autonomous path planning of the agriculture mobile robot is improved. By combining the man-machine cooperation path planning method based on the cloud model and the algorithm DGA*, the subjective initiative of people is fully exerted, the autonomous walking requirement of the agriculture mobile robot in a complicated dynamic environment is guaranteed, and the agriculture mobile robot has excellent operating performance for avoiding static and burst/dynamic obstacles.

The invention discloses an aero-engine body vibration characteristic analysis method. An engine body system is separated into a rotor system and a stator system, the rotor system and the stator are coupled through a fulcrum system and are simulated jointly through dynamic characteristics of the rotor system and the stator system; the engine body system is separated into the two study objects, i.e., a rotor bearing system and a stator bearing system, and then vibration response and dynamic bearing reaction of the rotor bearing system are simulated and analyzed through a finite element analysis platform under the action of unbalanced exciting force and rub-impact force to acquire excitory input of the stator system; at last, the acquired dynamic bearing reaction and the rub-impact force are loaded on the stator bearing system, and the vibration response characteristics of the stator bearing system is simulated and analyzed to acquire engine body vibration characteristics. The aero-engine body vibration characteristic analysis method has the advantages of reducing hidden fault danger in the design of aero-engines and providing decision bases and technical support for engine vibration faults.

The invention provides an image semantic annotation method based on super pixel segmentation. The method comprises the steps that a feature block extracted based on image super-pixel segmentation is input into a convolution neural network; a feature vector trained by the convolution neural network is expanded and weighted; and finally, a conditional random field model is constructed for semantic class annotation prediction. According to the technical scheme of the invention, the method takes a super-pixel block as a research object, which simplifies the complexity of the feature block extracted based on image super-pixel segmentation, and improves the calculation efficiency of semantic annotation; in addition, the multi-layer super-pixel block is used for semantic analysis, and the annotation results are integrated; and the accuracy and the robustness of semantic annotation are improved.

The invention discloses a mass concrete-based temperature field research method and a mass concrete-based temperature control method. The research method comprises the following steps of: establishinga mass concrete entity research object; measuring temperatures, in a pouring process, of a plurality of measurement points of the entity, and drawing corresponding temperature change curves to obtaintemperature field characteristics and a temperature change law; establishing a finite element analysis model according to the entity research object and automatically calculating temperatures of thecorresponding measurement points; adjusting temperature field calculation parameter of the model through comparing the calculated temperature values of the model with the actually measured temperaturevalues; and finally adjusting influence factors influencing hydration heat in a mass concrete model, carrying out analysis to obtain an influence result and an influence law, and adopting a corresponding temperature control method according to the result and the law. According to the mass concrete-based temperature field research method and the mass concrete-based temperature control method, fissures are prevented from occurring in mass concrete, and the pouring quality of concrete and the total quantity of projects are ensured.