93results about How to "Avoid preprocessing" patented technology

The invention discloses a multi-visual-field convolutional neural network-based image feature identification method. The method comprises the steps of collecting CT images with positive and negative tags in a historical database, and establishing a data set; judging a position region of a specified feature in each CT image of the data set by utilizing an image segmentation algorithm, and extracting sensitive regions of different pixel sizes; constructing a multi-visual-field convolutional neural network; inputting the extracted sensitive regions of different pixel sizes as samples to the multi-visual-field convolutional neural network, and training the multi-visual-field convolutional neural network to obtain a trained multi-visual-field convolutional neural network; and processing the to-be-identified CT images, inputting the extracted sensitive regions of different pixel sizes to the trained multi-visual-field convolutional neural network for performing feature identification, and determining the positive and negative tags of the to-be-identified images according to an identification result. According to the scheme, the end-to-end image identification is realized and the identification accuracy is ensured.

The invention discloses a vehicle classification method based on a convolution neural network. The vehicle classification method is concretely implemented according to the following steps that step 1: learning samples are acquired, and category tags are marked on the samples; step 2: the acquired learning samples act as training data to train a convolution neural network model so that excellent network model parameters are obtained; step 3: the characteristics of training data are extracted by using the trained convolution neural network model, and tenfold cross training is performed by using a support vector machine classification model constructed by a liblinear classifier so that the support vector machine classification model is obtained; and step 4: the characteristics of vehicle models under classification are extracted by suing the convolution neural network model, and then the vehicle categories to which vehicle images under classification belong are obtained by using the support vector machine classification model. All the connection layer output of the convolution neural network is used as the characteristic representation of the vehicle images and then the vehicle images are classified by using a SVM classifier so that the great accuracy of vehicle classification can be obtained.

The invention discloses a dynamic time sequence convolutional neural network-based license plate recognition method. The method comprises the following steps of: reading an original license plate image; carrying out license plate angle correction to obtain a to-be-recognized license plate image; inputting the to-be-recognized license plate image into a previously designed and trained convolutionalneural network so as to obtain a feature image and time sequence information, wherein the feature image comprises all the features of the license plate; and carrying out character recognition, inputting the feature image into a convolutional neural network of a long and short-term memory neural network layer on the basis of time sequence information of the last layer so as to obtain a classification result, and carrying out decoding by utilizing a CTC algorithm so as to obtain a final license plate character result. According to the method, vision modes are directly recognized from original images through using convolutional neural networks, self-learning and correction are carried out, the convolutional neural networks can be repeatedly used after being trained for one time, and the timeof single recognition is in a millisecond level, so that the method can be applied to the scenes needing to recognize license plates in real time. The dynamic time sequence-based long and short-termneural network layer is combined with CTC algorithm-based decoding, so that recognition error problems such as leak detection and repeated detection are effectively avoided, and the algorithm robustness is improved.

The invention provides a face diagnosis method and system based on deep learning. The method includes the steps of using a large number of human face images with disease labels for training a deep learning network, carrying iteration for multiple times, after parameters updating, the deep learning network being capable of automatically and effectively extracting the facial features including eyes,ears, mouths and eyebrows, and finding the internal relation between the facial features and various diseases of the human body so as to carry out effective disease detection and screen on the face images of new subjects. The method is a non-invasive computer-aided automatic diagnosis method. According to the system, the parameters can be updated by continuously collecting the facial images of patients and training, so that the prediction judgment result is more accurate. By means of the method and the system, the problem of difficulty in diseases inspection in poverty-trapped regions can beeffectively solved, so that people can conveniently and rapidly carry out non-invasive disease automatic detection and screening through a mobile phone, computer and other terminals, and the diseasescan be treated in time. The people's living quality is improved.

The invention discloses a method for reducing silicon carbide epitaxial basal plane dislocation density. The method comprises the steps of epitaxially growing a composite buffer layer of high-low doping concentration of multiple cycles on a SiC substrate, carrying out interface high temperature hydrogen etching processing on each single buffer layer, using interface high temperature processing and doping induction to introduce multiple interfaces, and using an interface image force to promote the conversion of a BPD defect to a TED defect. According to the method, the BPD defect in an epitaxial layer is greatly reduced, the BPD defect density in an epitaxial layer can be effectively reduced, the method is simple and the integration of epitaxial process is facilitated, at the same time the complex preprocessing of the SiC substrate is avoided, and the damage to the surface of the substrate is reduced.

The invention discloses a polycarbonate composite. The polycarbonate composite comprises, in parts by weight, 30-80 parts of polycarbonate, 5-40 parts of ABS (acrylonitrile butadiene styrene) and 5-80 parts of filler, wherein on the basis of the total weight of the polycarbonate composite, the weight content of copper elements is 0.01 ppm-20 ppm. A copper-containing compound with lower soluble copper content is added to the composite, and the content of the copper elements in the composite is controlled in the range of 0.01-20 ppm to reduce or avoid addition of other copper-containing compounds, pretreatment of filler powder can be avoided, and alkaline ions in a system can be stabilized, so that the prepared polycarbonate composite has remarkably improved forming toughness and forming use stability and further can combine the rigidity of the filler.

The invention provides a double-layer convolutional neural network-based license plate reconstruction method. The method includes the following steps that: blurring processing is performed on a clearlicense plate image on the basis of a random fuzzy kernel, so that a corresponding blurred license plate image can be generated; and blurred image blocks of a fixed size are intercepted from the blurred license plate image, and are inputted into a pre-designed de-blurring convolutional neural network, so that a de-blurred image feature layer is obtained; the same blurred license plate image blockis inputted into a pre-designed image enhancement convolutional neural network, so that an image enhancement mask set can be obtained; the de-blurred image feature layer and the image enhancement maskset are combined into a two-layer aggregation feature set, and a model is trained, so that reconstruction convolution parameters can be obtained; a blurred license plate image in an actual scene is inputted into a double-layer convolutional neural network, and is subjected to calculation with the reconstruction convolution parameters, so that a reconstructed license plate image can be obtained. With the method of the invention adopted, the quality of a blurred and degraded image can be improved; the contrast of the image can be improved; the sharpness of the image can be improved; and the edge and texture details of the image can be enhanced.

The invention relates to a method for quick synthesis of zinc oxide whiskers by means of microwaves. The method comprises the steps of 1, mixing, wherein zinc oxide powder and metal powder are taken in proportion and subjected to ball milling for 5-8 h, and then obtained powder is placed in an oven to be dried at 50-90 DEG C so that dried powder can be obtained; 2, microwave synthesis, wherein the dried powder obtained in step 1 is placed in a microwave high-temperature furnace for microwave synthesis, the heating rate of the microwave high-temperature furnace is 200-600 DEG C/min, heat insulation reaction is conducted for 1-5 min when temperature reaches 700-1200 DEG C, and then the zinc oxide whiskers are obtained. The process is simple, cost is low, operability is high, and the one-phase pure zinc oxide whiskers can be obtained at a low temperature within a short heat insulation time.

The invention provides application of a cholesterol molecule imprinted membrane sensing electrode in detecting cholesterol content in blood. Specific steps in the method are as follows: (1) treating a blood sample, that is, standing the blood sample at room temperature and carrying out centrifugation to extract serum if the blood sample is obtained by drawing blood from a person who is on an empty stomach, or carrying out anticoagulation by using heparin or EDTA (ethylene diamine tetraacetic acid) if the blood sample is blood plasma; (2) dropwise adding the blood sample obtained in step (1) onto the cholesterol molecule imprinted membrane sensing electrode, inserting the cholesterol molecule imprinted membrane sensing electrode into an amperometric sensor for determination, and recording readings. When used for detecting cholesterol content in blood, the molecule imprinted membrane sensing electrode has the advantages of low cost, high sensitivity, fast analysis speed, capacity of reducing sampling amount and avoiding tedious pretreatment of blood samples, etc.

The invention relates to a peanut nutritional milk and the preparation method, belonging to the technical field of food drink processing. The ingredients and the weight percentage of the invention are as follows: the protein 2.0 to 4.0 percent, sucrose 5.0 to 8.0 percent, emulsification stabilizing agent 0.15 to 0.25 percent, and remainder is the water. Compared with the prior art, the invention has the advantages that: (1) using the peanut butter protein directly to avoid the complicated crafts of pretreatment of peanut kernel, the protein extraction and other crafts, thus the production equipment disposition is simplified. (2) the nutritional milk prepared according to the invention has a protein content as high as 2.0 to 4.0 percent. (3) the nutritional milk presents cream color and has the unique fragrance and the taste of peanut and milk, taste is pure without burnt taste. During the guarantee of six months, the milk presents the uniform shape of the emulsion without lamination, flocculation phenomena and obvious deposit layer. Only by slightly shaking, the deposit can restore to the original condition.

The invention discloses a skew time series anomaly detection method based on a cost-sensitive hybrid network. The method comprises the following steps: firstly, establishing and training a cost-sensitive hybrid network model consisting of a deep convolutional neural network, a gated recursive network and a cost-sensitive loss function, wherein local features of a time sequence are learned througha deep convolutional neural network. Sequence characteristics of a time sequence are learned through a gated recursive network; then, the features are combined for classification; in the model training process, a cost sensitive loss function is used for measuring the similarity between an output result and a true value, then parameters of a network model are adjusted through a back propagation algorithm, and different penalty factors are used for punishing error detection of the network model for different numbers and types of samples. The method provided by the invention is simple, efficient,high in precision and relatively high in robustness, and has relatively high detection precision for both a skew time series data set and a non-skew time series data set.

The invention discloses a method for preparing nodular cast iron, which comprises the following steps: firstly determining chemical components and a mass ratio of each chemical component according the components and a mass ratio of each component of the nodular cast iron to be prepared so as to obtain raw materials, wherein the carbon element in the used raw materials is in a form of granular graphite, and the dimension grade equals the dimension grade of the graphite in the nodular cast iron to be prepared; then melting the raw materials in the presence of a protective medium or in vacuum environment, and cooling to obtain the nodular cast iron. The invention avoids the characteristics of excessive intermediate treatment processes and strict process parameter control in traditional preparation processes, overcomes the defects of low nodulizing grade and non-uniform spheroidal graphite distribution in melting processes of nodular cast iron, reaches nodularity of above 90%, and obtainsa nodular cast iron material with uniform distribution.

The invention relates to the field of catalyst synthesis, in particular to a preparation method for synthesizing a spherical-shell-type stable metal nano catalyst by a one-pot method. The catalyst prepared by the method is obtained mainly by the following steps: performing hydrolytic polymerization by using ethyl orthosilicate, resorcinol and formaldehyde as raw materials in an ethanol water reaction system under the catalysis action of ammonia water to form a spherical-shell-type material of phenolic resin coated silicon dioxide, directly adding a material salt as a metal precursor in the material preparation process, performing hydrothermal synthesis, performing centrifugal washing, performing filtration, performing drying to obtain a catalyst precursor, and performing carbon thermal reduction calcination at different temperatures in a nitrogen atmosphere to obtain the corresponding catalyst. Compared with a catalyst prepared by using single silicon dioxide or phenolic resin pelletsas a carrier or a conventional impregnation method, and the catalyst provided by the invention has a simple and convenient catalytic process, a better catalytic effect, more uniformly-distributed supported metal nanoparticles, smaller particle sizes, higher catalytic hydrogenation activity and improved catalyst stability.

The invention discloses a method for detecting volatile components in tobacco leaves. The method comprises the following steps: (1), mixing tobacco leaves with water, so as to obtain a mixed liquor; (2), suspending a solid phase microextraction device above the mixed liquor and carrying out headspace solid-phase micro-extraction, so as to obtain an extract liquor; during the process of headspace solid-phase micro-extraction, performing microwave processing on the mixed liquor, wherein the microwave power is 500 to 700W; (3), carrying out gas chromatography-mass spectrum association analysis, so as to obtain the detecting results of the volatile components in tobacco leaves. According to the method for detecting the volatile components in tobacco leaves, microwave-assisted headspace solid phase microextraction technology is used for analyzing volatile chemical components in tobacco leaves from different places of origin, the speed is fast, the efficiency is high, the sensitivity is high, and the analysis and detection results are comprehensive.

The invention discloses a multi-resolution volume rendering method based on intra-block interpolation. The multi-resolution volume rendering method based on the intra-block interpolation comprises the steps of data partitioning, blocking detail level selecting, secondary index establishing and volume rendering. The multi-resolution volume rendering method based on the intra-block interpolation has the beneficial effects that the introduction of extra data size and extra data preprocessing is avoided and the blocking effect of multi-resolution volume rendering is eliminated by use of intra-block border interpolation, and interpolation is completed by use of a GPU, and therefore, the volume rendering method is simpler and more efficient.

The invention relates to a radar clutter identification method based on a convolutional neural network. The method comprises the steps of obtaining clutter data; Dividing the clutter data into training data and test data; Constructing a convolutional neural network; Using the training data to train the convolutional neural network; Testing the trained convolutional neural network by using the testdata; And when the recognition accuracy of the convolutional neural network is greater than a preset accuracy, obtaining an optimal convolutional neural network. According to the method provided by the invention, the clutter data is processed and input into the convolutional neural network, and implicit feature extraction is carried out on the clutter data by utilizing the convolutional neural network, so that the feature extraction efficiency is improved; By reducing the number of clutter data, the sampling cost is reduced, and meanwhile, the recognition efficiency is improved; The convolutional neural network is used to classify and identify the clutter data, the identification accuracy rate reaches 99.51%, and the identification accuracy rate is greatly improved.

The invention relates to meat product adulteration detecting platform based on Raman spectrum. The meat product adulteration detecting platform comprises a sample pool (16), a light sensor (14), a controlled motor (15), a control module (13), a laser (1), a first focusing lens (2), a first light filter (3), a first one-way reflector (6), a second one-way reflector (4), a second focusing lens (5), a second light filter (7), a third focusing lens (8), a diffraction grating imaging system (9) and a computer, wherein the sample pool (16) comprises a sample pool body (16-4) and a transparent support plate (16-5) located in the sample pool body (16-4), a top cover (16-6) is disposed on the sample pool body (16-4), the sample pool (16) is provided with a sensor hole (16-3), the bottom of the sample pool (16) is provided with a measuring hole (16-2), and a focusing sleeve (16-1) is fixed at the measuring hole (16-2). By the meat product adulteration detecting platform, meat product adulteration detecting can be achieved fast.

The invention relates to a nisin separation and extraction method. The method comprises the following steps: blowing air to nisin fermentation liquor at a certain flow rate by adopting mixed gas of air and carbon dioxide at a certain volume ratio and a heated temperature so as to separate nisin from the foam fermentation liquor; collecting and defoaming to obtain a high purity nisin concentrated solution; adding a carrier and a stabilizer into the concentrated solution; and spraying and drying to prepare a nisin food preservative with qualified valence. High concentration macromolecular proteins such as thalli in the fermentation liquor subjected to hot air flows are easily separated from the fermentation liquor in form of precipitate and the precipitate is vacuum-dried to prepare thallus powder as nutritional feed. The method provided by the invention not only can be used for omitting labor-wasting, time-wasting and material-wasting high temperature pre-treatment requiring addition of massive hydrochloric acid solution into the fermentation liquor at present, but also avoids a process of filtering and collecting thallus residues by using a high cost ceramic membrane and filtering to remove thalli by using a diatomite sheet frame which generates a lot of solid residues, with high strength in a large site for a long time.

The invention provides an on-chip cell detecting system and method based on a PDMS chip. The system comprises an on-chip cell distinguishing and sorting device, and an injection pump, a gas pump, a tunable laser and a power detector which are connected with the on-chip cell distinguishing and sorting device. A sample outlet and a sample inlet of the on-chip cell distinguishing and sorting device are connected with the injection pump. A port of a micro-channel gas valve is connected with the gas pump. The input end and the output end of an optical fiber Bragg grating FP resonant cavity of the on-chip cell distinguishing and sorting device are connected with the tunable laser and the power detector respectively. The deformation of a PDMS thin film interlayer of the on-chip cell distinguishing and sorting device is controlled through the gas pump, and closing/opening is achieved. By means of the light detecting method, cell samples are not damaged, and the pretreatment such as dyeing or fluorescent marking of cell samples can be avoided.

The invention relates to restricted access poly(3-sulfopropyl methacrylate (SPM)-co-ethylene glycol dimethacrylate (EDMA)) modified silica gel chromatographic packing (Silica-g-p(SPM-co-EDMA)-g-pGMMA ([poly(glycerol monomethacrylate)]) and a preparation method thereof. The silica gel chromatographic packing is a material formed by grafting a double-layer polymer onto the surface of silica gel, and adopts poly(3-SPM-co-EDMA) as an inner layer and pGMMA as an outer layer. The expression of the silica gel chromatographic packing is Silica-g-p(SPM-co-EDMA)-g-pGMMA, wherein Silica is porous silica gel, g indicates grafting, p(SPM-co-EDMA) indicates a cross-linked polymer of 3-SPM and EDMA, and pGMMA indicates poly(glycerol monomethacrylate). The silica gel chromatographic packing can serve as a chromatographic stationary phase or a solid phase extraction material for analyzing biological samples and environmental samples, directly samples and extracts cationic compounds, and improves the analysis efficiency and accuracy. A method for measuring melamine and cyromazine in milk by online solid phase extraction/high-performance liquid chromatography coupling technology is provided and has good accuracy and reproducibility.

The invention belongs to a deep learning-based identifying and processing method for laser radar point cloud data. The result of a neural network model is computed by using a computing method of yolov2 when identification is carried out by using the trained neural network model; anchor values corresponding to the categories are utilized separately when the width and height of the result are computed; and average width and height of the same category of all objects in annotation data are preferred as the anchor values. According to the deep learning-based identifying and processing method, thepreprocessing process of the point cloud data is avoided; the categories of the objects can be identified; three-dimensional characteristics of the point cloud data are kept through converting the point cloud data into the forms of a distance matrix and a reflected light intensity matrix as input of a deep learning neural network, thereby ensuring relatively high identification rate on small objects and controlling the false identification rate at a very low level; and the finally obtained result is the result in a matrix dimension and can be converted according to the actual scene.

The invention discloses a human face expression intensity recognition method and system based on hidden variable analysis. According to the method, a twin convolutional neural network structure is adopted, face features extracted from the front end of a convolutional neural network are divided into orthogonal identity subspaces and expression subspaces through a hidden variable analysis method, the identity subspace feature difference of two branches of the twin network is minimized, the identity features and the expression features can be effectively separated; and meanwhile, sequencing constraint and semi-supervised regression training based on a time sequence are carried out on the expression subspace to obtain an expression intensity recognition model which can be used for continuous facial expression intensity recognition. According to the method, hidden variable analysis is used for expression intensity recognition, face features are divided into identity-related features and expression-related features, interference of identity information on expression intensity information is suppressed, and the robustness of expression intensity recognition can be effectively improved.

The invention discloses a trace heavy metal detecting system based on a chronopotentiometric stripping analysis method. The trace heavy metal detecting system comprises a chronopotentiometric stripping analysis device and a screen printing electrode, wherein the chronopotentiometric stripping analysis device comprises a housing, a working main board positioned in the housing, an electrode insertion groove, a built-in power supply, an input unit and an output unit, the electrode insertion groove, the built-in power supply, the input unit and the output unit adopt the housing as the platform andare respectively arranged on the housing, the screen printing electrode comprises a substrate, the substrate is printed with a working electrode, a counter electrode and a reference electrode, the three electrodes are respectively covered with an insulating layer, one ends of the three electrodes form and the opening holes arranged on the insulating layers form an electrolytic cell cavity, a sample to be detected is added into the electrolytic cell cavity in a dropwise manner, the screen printing electrode is inserted into the electrode insertion groove of the housing, and the reference electrode joint is connected to the constant potential module of the working main board. According to the present invention, the detection system has characteristics of simple composition and easy miniaturization, and is matched to the screen printing electrode, such that the electrode pretreatment is avoided, and the use is convenient.

The invention relates to the technical field of detection and in particular relates to a detection method for glycosylated hemoglobin and a kit used in the detection method. The method disclosed by the invention comprises the following steps: diluting a blood sample with water, and detecting by adopting a GlyHb chromatographic column through a high performance liquid chromatograph or a glycosylated hemoglobin instrument. According to the method, the blood sample can be directly fed to be detected after being diluted, so that the lengthy pre-treatment process can be avoided, and further the blood sample can be quickened to be detected in a quick and accurate manner. Tests prove that the method disclosed by the invention can be used for completing the quantitative detection on the glycosylated hemoglobin in the blood sample within 1min.