31results about How to "Improve the accuracy of discrimination" patented technology

This method combines the real-time filter with the call-bill analysis to filter the garbage notes (GN). It raises the intercepting rate of the GN supervision center and lowers the error judgment rate. Its advantages: 1) The Chinese key words adopt the blur-pinyin/ keyword matching mould-board technique to effectively clear the interfere info in GN. 2) According to the behavior character of GN, this method regulates dynamically the user trust levels, integrates the Garbage Degree Value of notes to judge the note genus. 3) The parallel real-time filter programs run on the multi-core hardware platform to reduce the influence of the note supervision center on the processing ability of the note business center. 4) The new strategy of note sorting and processing is: letting legal notes pass, intercepting GN, attaching runner's prompt on the question note, then letting it pass. 5) This system adopts Beyes method to learn off-line the note specimen and generates the GN keyword dictionary.

The invention discloses an equipment type identification and network intrusion detection method for the Internet of Things. the intrusion detection system is composed of an equipment type identification matching system and an abnormity detection system, and the type identification system can extract equipment characteristics according to characteristics of periodic communication of the equipment,classify and gather the equipment characteristics according to periodic statistical characteristics and the like, and classify the equipment into abstract types; The pattern recognition system based on the GRU neural network of the anomaly detection system can learn and memorize normal communication behaviors so as to establish a normal behavior sequence model, and due to the fact that the GRU neural network is designed for each device type, the recognition precision is higher, and the false alarm rate is greatly reduced; The gateway is adopted to monitor the communication condition of all Internet of Things devices in the network, so that all Internet of Things devices are directly or indirectly connected to the gateway, all communication between the Internet and the Internet and local communication between the Internet and the Internet of Things devices can be detected, local data processing is carried out through the gateway with higher performance, and shortage of resources of Internet of Things devices is avoided.

The invention discloses a sitting posture real-time detection method based on double cameras. The method comprises the following steps of calibrating internal and external parameters of cameras; acquiring a human body sitting posture stereo image through a binocular camera; performing stereo matching for the acquired stereo image, keeping a point cloud in a set range as a potential human body part point cloud; projecting the segmented human body part point cloud in the acquired image to perform image-based human face and human eye detection; judging whether the human faces are detected; judging whether an image rotation angle is greater than a set threshold value; judging whether the human eyes are detected; calling rotation transformation R2 of a calibrated camera coordinate system relative to a vertical plane, and calculating out vertical distance between the human eyes and the cameras through a three-dimensional point cloud nearby the human eyes; adding the set distance between the binocular camera and a desk surface and the distance between the human eyes and the cameras together to obtain the vertical distance between the human eyes and the desk surface; and judging whether the vertical distance between the human eyes and the desk surface is greater than a set vertical distance threshold value. The sitting posture real-time detection method based on double cameras automatically detects human body sitting posture in real time and judges whether the sitting posture is correct or not.

The invention discloses a Poa pratensis variety identification method based on a hyper-spectral image. The method comprises the following steps of: 1, placing Poa pratensis samples which are prepared in a hyper-spectral imaging device, and acquiring the hyper-spectral image; 2, extracting the information of the hyper-spectral image, which comprises spectral information and/or external characteristic information; 3, calculating total difference between different Poa pratensis varieties at different wave bands according to the spectral information which is extracted; 4, establishing a function model for determining different varieties of Poa pratensis according to the total difference at each wave band or according to the total difference and the external characteristic information; and 5, determining the samples according to the function model which is established in the step 4 to identify different varieties of Poa pratensis. The method provided by the invention is easy and convenient to operate, and has large technical advantages. The acquired information is accurate and reliable, and batch quick identification can be performed.

The embodiment of the invention provides a cattle illness data analysis method and device. The method concretely comprises the following steps: obtaining state index data representing cattle health states; selecting sample data, performing sample training of selected sample data according to an initialized cost matrix based on illness classification in advance and a multi-classification Bayesian process classifier based on cost sensibility, and obtaining the illness classification model of training samples; and according to the obtained state index data, the cost matrix and the training model, calculating the illness classification probability of the cattle to be detected and outputting the illness classification probability of the cattle to be detected. The method provided by the embodiment of the invention provides a cost-sensitive multi-classification Bayesian and Gaussian process to process the calculation of the illness classification probability of cattle, perform calculation of the state index states of the cattle and output a plurality of illness classification probabilities, and misjudgement cost factors are added to overcome the defects that a current discriminating method only can obtain general low misjudgement rate and only can output single classification results.

The invention relates to an otolith measurement and fish population discriminating method based on a polar coordinate, relates to the fish farming and management field and particularly relates to the method for distinguishing fish populations through measuring fish otolith. According to the method, a center point of an otolith picture is taken as a measurement reference point; end pointd, convex concave points or inflection points having prominent features in the otolith form are searched and taken as characteristic points; through establishing polar coordinates, the characteristic points and a core point are connected, radial length between the core point and each characteristic point is measured to acquire a set of characteristic length data; included angles of two adjacent connection lines are sequentially measured by utilizing the measurement software to acquire a set of characteristic angle data; on the basis of the characteristic length data and the characteristic angle data, a discrimination analysis method or a gray cluster analysis method is utilized, a population type of a detected sample is discriminated. Compared with a discriminating method through utilizing only the length, the otolith measurement and fish population discriminating method is advantaged in that the discrimination result is more accurate and reliable, and the fish population discrimination correct rate can be remarkably improved.

The invention discloses a domain name network address activeness statistics method and system. The method comprises the steps of step 1: cracking a domain name network address to obtain each sub-domain name; step 2: extracting a feature code from each sub-domain name; step 3: indexing all the sub-domain names by using the feature codes one by one to form a retrieval system; step 4: adding the feature codes of the sub-domain names into the retrieval system one by one and clustering all sub-domain names retrieved by a current sub-domain name into one category; and step 5: according to the sizes of all the sub-domain names clustered into one category and retrieved by the current sub-domain name, determining the activeness of the current sub-domain name. According to the method and system, through the analysis on a sub-domain name duplicate removal algorithm, it can be showed that a feature code based same-content sub-domain name removal and combination algorithm is a linear computing complexity algorithm and has very high judgment accuracy and recall rate.

The invention discloses a multi-sensor fusion-based driving safety guarantee system and method. The system comprises a main control board, a sensor group, a GSM communication module and a cloud data analysis platform. The main control board serves as a terminal of the cloud data analysis platform, uploads data to the cloud data analysis platform through the GSM communication module, and feeds backearly warning and evaluation performed by the cloud data analysis platform to a user; the GSM module finishes communication between the terminal and the cloud data analysis platform through a GSM; the sensor group comprises a plurality of sensors including a gas sensor, an alcohol sensor, a video sensor, a voice sensor and a six-axis sensor; and the cloud data analysis platform is connected withthe main control board and used for carrying out road illegal behavior analysis, traffic flow density information query, bad driving habit analysis and recommendation according to signals collected bythe sensor group.

The invention discloses a human body behavior recognition method based on a multi-class spectrogram and a composite convolutional neural network. The method comprises the steps: sampling different human body behaviors through a stepping frequency continuous wave radar, and obtaining a matrix in each sampling; selecting a plurality of column vectors for each matrix, and constructing a spectrogram sample data set by adopting three different time-frequency analysis methods; constructing a composite convolutional neural network and training to obtain a classification model. According to the method, multiple types of spectrograms with different feature expressions are used as input data, so that expressions about human behavior features are diversified and comprehensive; when the composite convolutional neural network is used for feature extraction, independent spatial features in a single spectrogram and associated features among different spectrograms can be fully extracted, so that feature extraction is sufficient and effective; the two features are fused and then sent to the classifier for classification, so that the classifier can fully utilize different types of features, and the discrimination accuracy is improved.

The invention discloses a fluorescence characteristic standard spectrum library for identifying paralytic shellfish poisoning-producing microalgae as well as a construction method and application of the fluorescence characteristic standard spectrum library. The method comprises the following steps of: extracting three-dimensional fluorescence spectrum information of PSP-producing algae and non-PSP-producing algae growing under different environmental conditions, extracting characteristic peaks of experimental algae by using a Db7 wavelet function, performing clustering analysis by using a system clustering method, removing abnormal spectrums, and screening standard spectrums to obtain a Db7 fluorescence characteristic standard spectrum library. When a discriminating function established according to the fluorescence characteristic standard spectrum library is used for discriminating the PSP-producing algae and the non-PSP-producing algae, the discriminating correct rates are 84.6% and95.6% respectively, the correct rates are very high, and the purpose of quickly and accurately identifying the poisoning-producing algae is basically achieved. The fluorescence characteristic standardspectrum library can be used for preparing a red tide algae identification sensor or a portable algae fluorescence identification instrument.

The embodiment of the invention provides an interface display method of nuclear power equipment parameter information. Device and electronic equipment, The method comprises the following steps of: obtaining a sample; Parameter information of the nuclear power equipment is acquired; when the parameter information accords with a preset condition; performing first classification processing on the parameter information to obtain first class information of the parameter information, performing abnormity judgment on the parameter information according to a processing strategy determined by the firstclass information to obtain a judgment result, According to the method, the parameter information of the nuclear power equipment is acquired, the display mode of the parameter information is determined according to the judgment result, the parameter information is displayed according to the display mode, whether the parameter information of the nuclear power equipment is abnormal or not can be quickly and effectively judged, and judgment accuracy is improved; The display mode of the parameter information is determined according to the abnormal judgment result of the parameter information, andthe recognition efficiency can be improved.

The invention discloses a homonymous cell distinguishing method and system based on text similarity. The method comprises the following steps: acquiring a first name of a first cell to be distinguished and a second name of a second cell to be distinguished; acquiring first attribute information of the first cell to be distinguished and second attribute information of the second cell to be distinguished; when the first basic information is the same as the second basic information, determining the distance between the first to-be-distinguished cell and the second to-be-distinguished cell according to the first longitude and latitude information and the second longitude and latitude information; when the distance is smaller than or equal to a preset threshold value, calculating text similarity between the first name and the second name; and determining a discrimination result according to the text similarity and the distance between the two cells. Due to the fact that the distance betweenthe to-be-distinguished cells and the text similarity between the cell names are comprehensively considered, misjudgment caused by alias of the to-be-distinguished cells or the same name of the to-be-distinguished cells can be avoided, and distinguishing accuracy is effectively improved.

The invention provides a training method of a target detection model for multi-category junk scene recognition. The training method comprises the following steps: constructing a basic target detection model; in the p convolution layers, selecting m continuous convolution layers, and replacing each selected convolution layer by a cavity convolution; sequentially setting the void ratios of the m hole convolutions to meet the condition that the largest common divisor of the void ratios of any two adjacent hole convolutions is 1 so as to obtain an optimized target detection model; and training the optimized target detection model by using the sample image training set to obtain a target detection model for multi-category junk scene recognition. According to the invention, the problem of low garbage target detection precision of a scene image with a wide visual field due to a small receptive field range of a target detection model in the prior art is solved, the scene image is used as a model input for garbage recognition detection, and the obtained detection result often has the phenomenon of invalid multi-point garbage identification or false alarm of garbage detection.

The invention discloses a multi-feature fusion homonymous cell identification method and system. The method comprises the following steps: acquiring first/second longitude and latitude information offirst/second cells to be identified; determining the distance between the two cells; when the distance is smaller than or equal to a first preset threshold value, obtaining first/second images and first/second names of the first/second cells to be identified; calculating first/second LBP feature values of the first/second images, and determining an image similarity value between the first image and the second image; calculating a text similarity value between the first name and the second name; according to a preset weight, performing weighted average on the image similarity value and the textsimilarity value to obtain a similarity score between the two cells; and determining a discrimination result according to the similarity score. According to the method, the three characteristics of the distance between the to-be-distinguished cells, the image similarity value and the text similarity between the cell names are comprehensively considered, so that misjudgment can be avoided, and thedistinguishing accuracy is effectively improved.

The embodiment of the invention provides a shared internet access detection method and device and a storage medium. The method comprises the steps that a TCP flow table is established according to uplink TCP SYN messages sent by terminal equipment, a plurality of flow information identifiers are recorded in the TCP flow table, and each flow information identifier is used for representing the corresponding uplink TCP SYN message; decoding information and a TCP timestamp value corresponding to the uplink TCP SYN message are acquired according to the TCP flow table; the received uplink TCP SYN messages are grouped according to the decoding information; classification information of each group of uplink TCP SYN messages is acquired, and the classification information represents a classification result of classification of each group of uplink TCP SYN messages according to the corresponding TCP timestamp value; and according to the classification information, judging whether the terminal equipment has a shared internet access behavior or not.