39results about How to "Reveal well" patented technology

The invention discloses a real-time expression recognition method based on a multichannel parallel convolutional neural network (MPCNN). The method comprises the steps of: extracting expression data, which contains RGB and Depth images, from a facial-expression data set; carrying out local binarization preprocessing (LBP) and facial-key-point extraction preprocessing on the color images, carrying out gradient preprocessing on the depth images, dividing the preprocessed images into two parts of a training set and a test set, and building the multichannel parallel convolutional neural network; sending the preprocessed images in the training set to the network for training to obtain depth channel, lbp channel and key-point channel recognition models which learn facial-expression outlines, stereoscopic distribution and key-point features; and adopting maximum confidence to fuse classification results of the three recognition models to obtain a final expression recognition model, and building a real-time expression recognition system. The method enhances the robustness of the recognition network, and effectively improves the performance of the real-time expression recognition system.

The invention provides a breast tumor diagnosis system based on magnetic resonance spectrum imaging, comprising a superconduction MR (Magneto Resistance) scanner and a computer system. Through a computer auxiliary detection measure, the breast tumor diagnosis system realizes the early identification diagnosis of a breast tumor and the further diagnosis of breast cancer. Mammary gland magnetic resonance spectrum data are studied by adopting a distinctive manifold study method and projected to a low dimension embedding space, thus not only a low dimension manifold structure hidden in a high dimension magnetic resonance spectrum space can be revealed, but also the identification information in the mammary gland magnetic resonance spectrum data can be efficiently kept; then optimization clustering is carried out on the low dimension identification characteristic by utilizing a clustering method so that data points without the similar identification characteristic are separated to the maximum; and further a cost sensitive mechanism is introduced to achieve misclassification total cost minimization and realize optimization diagnosis of the breast cancer.

A neural network wind power forecasting method and system includes: collecting numerical weather forecasting data at forecasting time; the wind power prediction value is obtained by substituting the numerical weather forecast data into the prediction model constructed in advance. The prediction model is based on principal component analysis and neural network. The technical scheme of the inventioneffectively improves the prediction accuracy, shows that the method has certain feasibility and advancement in wind power prediction based on numerical weather prediction, and has certain advantagesin processing large sample data.

The invention discloses a communication signal modulation mode identification method, which overcomes the problems of low identification rate and limited identification types under a low signal-to-noise ratio, and comprises the following steps of: 1) forming a set of known signals: (1) segmenting the known signals; (2) performing time-frequency conversion on the N sections of signal data; (3) performing gray scale transformation on the N three-dimensional time-frequency graphs; (4) marking the N two-dimensional time-frequency graphs; 2) training a convolutional neural network: (1) setting theconvolutional neural network; (2) training the set convolutional neural network; (3) adjusting parameters of the convolutional neural network; (4) generating a trained neural network; 3) forming a setby the received unknown signals; and 4) identifying a signal modulation mode, wherein the convolutional neural network is formed by training according to the set formed by the known signals, and theset formed by the unknown signals is input into the convolutional neural network for recognition of the modulation mode to obtain the modulation mode of the signals.

The invention discloses a triaxial multiple coal and gas outburst test method. The method comprises the following steps of: 1) installing a telescopic steel sleeve provided with a raw coal test specimen between a confining pressure cavity upper cover and a lower bearing plate; 2) placing various sensors in the telescopic steel sleeve; 3) arranging a sealing aluminium sheet I on a sealing taper sleeve; 4) installing a primary pressure chamber and a secondary pressure chamber at the top of a main pressure chamber; 5) monitoring variations of stresses, temperatures and gas pressures of the raw coal test specimen before and after tests; 6) pressurizing the test specimen in the main pressure chamber; 7) evacuating air inside a rubber sleeve of the main pressure chamber; 8) filling gases into the raw coal test specimen in the main pressure chamber and the secondary pressure chamber; 9) carrying out a primary outburst test; and 10) carrying out a secondary outburst test. By adopting the test method, raw coal in different sizes can be adopted to carry out coal and gas outburst tests in the triaxial stress state to study the coal and gas outburst occurrence principles more comprehensively and scientifically.

The invention relates to a paddy rice automatic change detection method integrating pixels and objects. The method comprises the following steps of establishing an MODIS-based water body index time series data set and MODIS water body index time series curves; calculating the similarity of the MODIS water body index time series curves in the starting and ending years pixel by pixel; judging a suspected change region based on the similarity of the MODIS water body index time series curves; obtaining an image object through image segmentation by integrating multi-phase Landsat images in the starting year; superposing the suspected change region with the image object to obtain a suspected change object of a research region; further subdividing the suspected change object of the research region by integrating multi-phase Landsat images in the ending year; and performing paddy rice charting of the starting and ending years on the subdivided suspected change object of the research region by utilizing Landsat remote sensing image characteristics so as to obtain a paddy rice dynamic change distribution diagram of the research region in the starting and ending years. The method has the characteristics of high automation degree, simplicity, easiness in use, good robustness, high classification precision and the like.

The invention discloses a CELMDAN and SSKFDA-based mechanical state monitoring method. The CELMDAN and SSKFDA-based mechanical state monitoring method includes the following steps that: (1) a CELMDANmethod is adopted to decompose complex vibration signals into a plurality of product functions with physical significance; (2) a method of taking a periodic modulation intensity PMI as a PFs selectioncriterion is provided, so that effective PFs can be accurately selected. (3) an SSKFDA dimension reduction method is provided, geometrical information of a label sample and an unlabeled sample set isfully utilized, a kernel method, sparse representation, manifold learning and an FDA method are fused, a low-dimensional subspace data set embedded in a high-dimensional sparse space is better disclosed, and the problem of dimension reduction of high-dimensional, sparse and nonlinear data is solved. (4) a rapid SSKFDA model selection method is proposed, according to the method, optimal model parameters are solved based on the criterion of minimum intra-class local structure measurement and maximum full-local structure measurement. and (5) a mechanical known and unknown state detection methodbased on global monitoring statistics and Bayesian posteriori reasoning is proposed, and the problem that most mechanical monitoring systems cannot detect unknown abnormal states is well solved.

The invention belongs to technical fields of medical measurement, computer simulation and computational fluid mechanics and discloses a heart left ventricle flow field domain numerical simulation method, a computer and a computer program. Contour extraction is performed according to an echocardiogram-based left ventricle contour extraction method to realize left ventricle blood flow field dynamicsimulation; a left ventricle initial contour is extracted on the basis of echocardiogram, contour pixel point coordinates are acquired according to a quick connected domain method, and a left ventricle grid model is automatically generated on the basis of an extracted heart left ventricle contour model; optimal boundary conditions are set, and a heart left ventricle flow field change process is dynamically simulated. Cardiovascular disease occurrence and development pathophysiology and pathological anatomy foundation can be better revealed, and valuable reference information can be provided for prevention and treatment of the cardiovascular disease.

The invention discloses a triaxial one-time coal and gas outburst test method. The method comprises the following steps of: 1) installing a telescopic steel sleeve provided with a raw coal test specimen between a confining pressure cavity upper cover and a lower bearing plate; 2) placing various sensors in the telescopic steel sleeve; 3) arranging a sealing aluminium sheet I on a sealing taper sleeve; 4) installing a primary pressure chamber at the top of a main pressure chamber; 5) monitoring variations of stresses, temperatures and gas pressures of the raw coal test specimen before and after tests; 6) pressurizing the test specimen in the main pressure chamber; 7) evacuating air inside a rubber sleeve of the main pressure chamber; 8) filling gases into the raw coal test specimen in the main pressure chamber and the primary pressure chamber; and 9) piercing the sealing aluminium sheet I with a piercing tool, and enabling the raw coal test specimen in the rubber sleeve and the gases to carry out one-time outburst. By adopting the test method, raw coal in different sizes can be adopted to carry out coal and gas outburst test in the triaxial stress state to study the coal and gas outburst occurrence principles more comprehensively and scientifically.

The invention provides a coauthorship network dividing method for selecting a seed node according to academic association centrality, relates to the fields of academic author collaborated writing behavior analysis and association dividing, and particularly to a community dividing algorithm which can display a community structure in a complicated coauthorship network. The method comprises the following steps of firstly, according to different attributes of an article which is issued by authors, calculating a collaboration strength by means of a Salton method and constructing the coauthorship network; secondly, considering self impact degrees of the authors and influences of adjacent scholars to the authors, and calculating the academic association centrality of the node; and finally, using the academic association centrality of the node as a seed node selection basis, combining the academic association centrality into a second period of a Louvain community dividing algorithm, namely preventing merging between the seed nodes, and merging the non-seed nodes into the community to which the seed node belongs as possible. The coauthorship network dividing method settles problems of excessive merging between middle-sized communities and large communities and a large number of small communities, and can more accurately find out and divided the academic association in the coauthorship network.

The invention discloses a gait-electrocardiogram (ECG) RR interval correlation method based on Gaussian regression. The invention comprises the following steps: step 1, synchronously acquiring three-dimensional coordinate data of motions of a subject and corresponding ECG signals by using a three-dimensional motion tracking system and an ECG acquisition device, respectively; step 2, carrying out gait quantitative analysis, nonlinear analysis and time-frequency analysis on the three-dimensional coordinate data, extracting 17 features to form a feature vector, and providing a fusion method; step 3, pre-processing the ECG signals to extract RR intervals; step 4, constructing a Gaussian process regression (GPR) correlation prediction model between the gait features and the ECG RR intervals; and 5, training and testing the GPR model, and performing correlation prediction analysis. The invention has advantages of strong self-adaptation, easiness in realization, super-parameter self-adaptive acquisition, etc., has better robustness and generalization performance, and can better reveal correlation information between the gait and the ECG.

The embodiment of the invention provides a risk phrase recognition method and device, electronic equipment and a storage medium, and relates to the technical field of text processing. The method comprises the following steps: carrying out phrase recognition on a risk description text by adopting a predetermined phrase recognition algorithm to obtain a first risk phrase list; processing the risk description text by adopting a predetermined word segmentation tool to obtain a second risk phrase list; then, merging the first risk phrase list and the second risk phrase list to determine a risk phrase list including a plurality of risk phrases. According to the method provided by the embodiment of the invention, the risk phrases can be quickly and accurately identified, the identified phrases are more comprehensive, the information amount is larger, and the content of the risk theme can be well revealed.

The invention discloses an individualized brain function map construction method and system, and relates to the image processing field, the method comprises the following steps: acquiring a scanning image of a user; calculating the functional connection between each voxel and the whole brain voxel in the interested brain region of the scanned image to obtain a functional connection matrix; and calculating a similarity matrix of any two voxels according to the function connection matrix, performing similarity clustering on data in the similarity matrix to obtain a plurality of sub-regions, andperforming brain region division on the user according to the sub-regions in combination with the scanning image to construct an individualized brain function map. By constructing the individualized brain function map, the brain function region of the individual can be positioned more accurately, so that the neural mechanism of the brain function of the individual can be better revealed, an individualized diagnosis and treatment scheme can be conveniently made, and nerve regulation, rehabilitation and the like can be better guided.

The present invention discloses a fiber Bragg grating sensing and infrared sensing technology-based multi-physical field measurement system. The system comprises a fiber Bragg grating sensing unit, an infrared thermal imager sensing unit, a data acquisition unit and a control processing unit; the fiber Bragg grating sensing unit comprises a fiber Bragg grating sensor and an optical fiber demodulator which are connected with each other, wherein the fiber Bragg grating sensor is attached to a workpiece to be measured, and the optical fiber demodulator is connected with the control processing unit; the infrared thermal imager sensing unit comprises an infrared thermal imager and a digital image acquisition card which are connected with each other, wherein the digital image acquisition card is connected with the control processing unit; the data acquisition unit includes a digital acquisition card with an input and output function, wherein the output end of the digital acquisition card is connected with the optical fiber demodulator and the infrared thermal imager, and the input end of the digital acquisition card is connected with the control processing unit; and the control processing unit includes an industrial personal computer and a software platform. With the fiber Bragg grating sensing and infrared sensing technology-based multi-physical field measurement system of the invention adopted, precision measurement of various kinds of physical fields in the vicinity of a cutting area during a machining process can be realized, and the research of a cutting mechanism is facilitated.

The invention discloses a blind domain image sample classification method based on an over-limit hidden feature model, and the method employs a cross-domain data generation and augmentation strategy,and efficiently reduces the deviation between different data domains. Specifically, the over-limit hidden feature model incorporates an ELM and ELM-AE (an automatic encoder based on ELM) into a unified optimization model, and advantages of the ELM and ELM-AE are inherited, and good image classification and image reconstruction effects can be obtained under the condition that original data information is protected, the image classification capability and the image reconstruction capability are achieved, the potential relation between the original image data and high-level semantics can be better disclosed, information loss is reduced, and the image classification precision is improved. The blind domain adaptation problem is solved by using the over-limit hidden feature model, cross-domain knowledge is fused, domain deviation between different domain data is effectively reduced, and the blind domain data classification accuracy is improved.

The invention discloses a fingerprint spectrum construction method and a detection method of a Yangjing Zhongyu prescription. The construction method comprises the following steps: preparing a test solution; preparing a reference solution; and constructing a fingerprint spectrum: sucking the test solution and the reference solution, and injecting the sucked test solution and the sucked reference solution into a high performance liquid chromatograph to be measured so as to obtain the fingerprint spectrum of the Yangjing Zhongyu prescription with a common characteristic peak. On the basis of long-term experience accumulation and a large number of experiments, the inventors select ferulic acid, paeoniflorin, morroniside and loganin as reference substances to construct the fingerprint spectrum, so that the quality of the classical Yangjing Zhongyu decoction can be well characterized, and the quality of the modern Yangjing Zhongyu condensed decoction and Yangjing Zhongyu granules can also be characterized. Especially, the fingerprint spectrum is constructed by selecting appropriate chromatographic conditions, so that the Yangjing Zhongyu decoction can be qualitatively and quantitativelycharacterized, and the Yangjing Zhongyu condensed decoction and the Yangjing Zhongyu granules can also be qualitatively and quantitatively characterized.

The invention discloses a stepless dynamic evolution theme cloud picture generation method based on automatic indexing. Comprising the following steps: setting corresponding subject terms for a certain field, and screening and indexing the subject terms in all texts in the field; according to the release time sequence of the text, counting the statistical state of the subject terms changing along with time in the field; displaying the statistical state of the subject term in the subject term cloud picture through the states such as the font size and the color of the subject term; realizing stepless change of dynamic display of the theme cloud picture by matching a time axis with a key frame in animation transformation. According to the invention, the theme cloud picture display items with higher quality are generated, the number of the theme word display items used for visualization is smaller, the theme word display items have higher field representativeness, meaning superposition among the theme words and meaningless display items in the field can be reduced, and theme content can be better revealed; and the development condition and change of each subject field can be disclosed more scientifically and carefully.

The invention discloses a biological edge mark system and an establishing method thereof, which can simply and efficiently find out a key interactive change as a biological marker of disease occurrence and development. The technical scheme provided by the invention is as follows: the establishing method of the biological edge mark system comprises the steps: collecting data with double states; selecting gene pairs, the correlation of which meets significant difference conditions; for the gene pairs, the correlation of which meets the significant difference conditions, translating expression value data of the gene pairs into edge data representing the correlation through matrix transformation; applying a feature selection algorithm to find out a gene pair with the optimal classification capacity from the edge data, and utilizing the gene pair with the optimal classification capacity as a biological edge mark, thereby establishing a biological edge mark system.