260results about How to "Improve the detection effect" patented technology

The invention provides a real-time human body detection method based on an AdaBoost frame and colour of a head, comprising training and detecting steps: (1) extracting multi-scale HOG (histograms of oriented gradients) characteristics according to a template; (2) training a human body detection model by adopting an AdaBoost-Boosting method; (3) extracting histogram characteristics of the colour of the head; (4) training a head discrimination model by adopting an AdaBoost method; (5) detecting a human body based on a sliding window method; (6) for each detection window, firstly extracting the HOG characteristics, and judging whether the detected object is a human body or not according the human body detection model; and (7) for the window which is determined to be a human body, extracting the histogram characteristics of the head, judging whether the head is contained or not; and determining the window containing the head to be a window containing a human body, and drawing a rectangle at a corresponding position in an image. In the invention, the characteristic unit, namely a Block of the original HOG characteristics is adopted, the Block is of multi-scale type, and a characteristic template is combined, thus the detection effect is improved; and the head characteristic discrimination is added, thus the detection rate is improved.

The invention discloses a mercapto-beta-cyclodextrin modified silver nano-rod array, its preparation method and its use. The array is a silver nano-rod array with the surface modified by mercapto-beta-cyclodextrin, wherein the lengths of silver nano-rods are 150-600nm, the diameters of the silver nano-rods are 60-90nm, the space between each two silver nano-rods is 10-40nm, and the substrate of the array is formed by superposing one or more than two of a silver film, a gold film, a copper film and a nickel film which have the thicknesses of 20-50mum. The preparation method comprises the following steps: obtaining a through hole alumina template through using a secondary anode oxidation method, sputtering a metal conductive film on one surface of the through hole alumina template through using an ion sputtering method to obtain an alumina template with the metal conductive film, carrying out electric deposition through placing the alumina template with the metal conductive film in a silver electrolyte to obtain an alumina template with holes provided with the silver nano-rods and one surface coated with the metal conductive film, etching off the alumina template through putting the resultant alumina template in an acidic or strongly alkaline solution, and immersing the resultant material in an aqueous solution of the mercapto-beta-cyclodextrin for at least 1h to prepare the mercapto-beta-cyclodextrin modified silver nano-rod array. The mercapto-beta-cyclodextrin modified silver nano-rod array can be used for rapidly detecting trace PCB77 or PCB101.

The invention discloses a fiber-Raman-enhanced probe based on a tapered optical fiber and a micro-ball lens and a manufacturing method of the fiber-Raman-enhanced probe. The fiber-Raman-enhanced probe comprises the optical fiber with a tapered end, a micrometer-scale ball and glue connecting the tip of the tapered optical fiber and the ball. Single-color laser enters the optical fiber from one end of the optical fiber in a coupling manner, reaches the tapered end of the optical fiber, and converges to the surface of a sample through the ball lens at the tip of the tapered optical fiber. The probe can focus detecting light into a thin and long light beam, the width of the focused light beam is the size of sub-wavelength, and the length of the focused light beam is about three times of the wavelength. When the probe is used as the Raman probe, the thin and long focused light beam can stimulate more to-be-detected molecules, and the sub-wavelength width can increase the spatial resolution of detecting. Meanwhile, due to the fact that the tapered optical fiber is coupled into the micro-ball and then focused to the sample, the irradiating area of excitation light is reduced, only a focus point has the enhanced excitation light, and the spatial resolution and signal-to-noise ratio of Raman detecting are increased favorably.

The invention provides a structural damage assessment method based on an ultrasonic guided-wave drive/sensing array. The structural damage assessment method comprises the following steps: firstly, pre-arranging or bonding ultrasonic guided-wave sensing units on the surface of a structure to be tested to form an ultrasonic guided-wave sensing array, exciting the various ultrasonic guided-wave sensing units in turns to generate ultrasonic guided-wave signals, and receiving the reflected guided-wave signals of the internal damages/boundaries of the structure to be tested through the ultrasonic guided-wave sensing array; then carrying out chromatic dispersion removal processing on the reflected guided-wave signals, globally imaging the structure to be tested by virtue of a phased array imaging algorithm, and locating a damage area; finally, carrying out accurate quantitative analysis on damages in the damage area by virtue of a reverse time migration imaging technology based on a multiple signal classification method with regard to the damage area and on the basis of the reflected guided-wave signals. The method realizes rapid and accurate location and assessment on damages in large structures, the exploration capacity is greatly improved, and the needed labour and maintenance expense are also reduced.

This invention provides a system and a method for monitoring air targets based on public mobile communication network, which sets up or reforms a communication base station as the air transmission base station and distributes a certain numbers of passive receiving stations around it. The passive receive stations receive the echoes of the air transmission base station radiation signals reflected by the air target and measure the delay time, arriving angle, frequency and phase variance of the echo then to transfer the data to a data process center via network, which processes the data in terms of multi-curve surface intersection method to get data of location, speed, course of the air target and send them to users.

The invention discloses a kit for simultaneously detecting nucleic acids of seven respiratory pathogens and application of the kit. The kit is based on a double amplification technology (RNA isothermal amplification and multi-biotin signal amplification), and can detect H1N1/H3N2 type influenza A viruses, influenza B viruses, respiratory syncytial viruses, 1/2/3 type human parainfluenza viruses, B/E type adenoviruses, mycoplasma pneumoniae and chlamydia pneumoniae. The kit of the invention does not need RNA extraction, and is not prone to pollution, high in sensitivity and good in specificityin the process of detection, and can be widely applied to detection of the nucleic acids of the above seven respiratory pathogens.

The invention relates to an automatic voice summation tone detection method based on a deep neural network. The method comprises a step of training a deep neural network model for summation tone detection, wherein an input layer of the deep neural network model is characteristic information of voice and an output layer is probability output values for three states which are summation tone voice, voice of a single person and non-speech sound. The depth neural network model is used to perform summation tone detection on the automatic voice.

The invention discloses a face anti-counterfeiting detection method of a double-channel convolutional neural network based on an attention model, and the method comprises the steps: constructing a training set which comprises a training sample composed of an RGB face image and an MSR face image; constructing a face anti-counterfeiting detection network which comprises a feature extraction unit used for extracting RGB feature vectors and MSR feature vectors, a feature fusion unit used for fusing the RGB feature vectors and the MSR feature vectors, and a feature classification unit used for classifying the fused feature vectors; training the face anti-counterfeiting detection network by using the training set to obtain a face anti-counterfeiting detection model; during application, preprocessing a to-be-detected RGB face picture and then inputting the preprocessed face image into the face anti-counterfeiting detection model, and outputting a detection result, namely a true face or a false face, through calculation. According to the method, the influence of illumination on detection and rich texture information of the RGB picture can be reduced by using the MSR picture at the same time, so that the detection result is accurate and has certain generalization.

The invention provides a bad webpage detection method and device. The bed webpage detection method comprises the following steps of: acquiring suspected bad webpages corresponding to bad keywords according to the bad keywords, acquiring an original address corresponding to each suspected bad webpage and generating a bed website list containing the original addresses; analyzing the suspected bad webpages to generate an analysis result; and when vicious skips existing in the bad webpages are identified according to the analysis result, deleting the original addresses corresponding to the suspected bad webpages from the bad website list. According to the bad webpage detection method and device provided by the invention, because the suspected bad webpages acquired according to the bad keywords are further analyzed, the webpages with vicious skips are deleted, and the detection effect on the bad webpages is improved.

The invention discloses an android malicious code detection system and method based on Opcode backtracking. The method includes the steps that firstly, a rule base is established, and detection strategies are customized by the rule base according to an object to be detected; an API called initially is placed in a queue as a name to be matched; the name to be matched is read from the queue, and names of functions called by all Opcode instructions are obtained by using a dex analysis and Opcode disassembling module; the names of the called functions are contrasted with the name to be matched, the name of the function where the successfully matched Opcode instruction belongs is placed in the queue as a name to be matched, the operations are repeated until the queue is empty, whether a called relational tree conforms to detection rules customized in the rule base or not is analyzed and judged on the basis of the rule base, and if yes, the relational tree is a malicious sample; otherwise the relational tree is safe. According to the Android malicious mode detection system and method based on Opcode backtracking, the problems of false alarm and alarm failure in an existing malicious code detection method can be well solved.

The invention discloses a nanometer annular cavity SERS substrate based on a surface plasmon effect. The nanometer annular cavity SERS substrate comprises a substrate, wherein a nanoscale single layer PS microsphere array is arranged on the substrate; the PS microsphere array is filled with a gel layer formed by SiO2 or TiO2 precursor solution; the thickness of the gel layer is smaller than the diameter of microspheres; a nanoscale annular concave cavity is formed between the top of the PS microspheres and the gel layer; a metal layer is coated on the surface layer of the substrate; a metal annular cavity array is formed in the annular concave cavity coated with the metal layer. A to-be-detected sample is filled in the annular concave cavity, a cylindrical surface plasmon is formed in the metal annular cavity during illumination, intensive local electric field enhancement is formed, a Raman signal which can be detected is excited by the to-be-detected sample by virtue of the enhanced electric field, and the nanometer annular cavity SERS substrate is simple in structure and easy to process and prepare. The invention also discloses a method for manufacturing the SERS substrate. An annular cavity array is prepared on a planar substrate by adopting nanometer microspheres and a spin-coating method. The method is simple and easy for large-scale production.

The invention discloses an assessment device for a fillet weld of an end flange of a GIS (gas insulated switchgear) and a barrel and a lossless assessment method, aiming at solving the problem that an existing detection method cannot effectively achieve lossless assessment on the fillet weld of the end flange of the GIS and the barrel. The device is characterized in that an ultrasonic probe is installed on a wedge block, both the ultrasonic probe and a encoder are connected with a scanning rack so as to run synchronously, the encoder is connected with an ultrasonic wave phased array detector through a data line, the ultrasonic probe is connected with the ultrasonic wave phased array detector through the data line, and the surface of the ultrasonic probe is coated with a coupling agent and is connected with the wedge block through bolts. The assessment device can realize lossless assessment of total cross section of the structure, can detect both surface defect and interior defect, can detect both the outer fillet weld of the end flange of the GIS in service and the barrel and the inner fillet weld, and can comprehensively assess the quality of the structure.

The invention provides a coarse-to-fine fingerprint identification method fusing second-level and third-level features. The method includes the following steps that firstly, a high-resolution fingerprint image database comprising the third-level fingerprint features is constructed; secondly, fingerprint minutiae features and the third-level features in a collected fingerprint image are calibrated; thirdly, whether the number of the minutiae features included in the collected fingerprint image is smaller than the threshold or not is judged, if yes, the step e is executed, and if not, the step d is executed; fourthly, the minutiae features are used for performing primary retrieving and matching, and a first candidate fingerprint set with a high matching score is solved; fifthly, accurate matching is performed with the third-level features, and a second candidate fingerprint set with a high matching score is solved; sixthly, the collected fingerprint image and fingerprints in the second candidate fingerprint set are determined. According to the coarse-to-fine fingerprint identification method fusing the second-level and third-level features, under the premise that efficiency is taken into consideration, the fingerprint retrieving and matching precision is achieved with high precision.

The invention provides a railway foreign matter invasion detection method based on railway monitoring. The method comprises the following steps: training YOLOv3 by utilizing an original image containing foreign matter invasion of a monitoring video in a known railway scene; obtaining a trained model weight; acquiring a monitoring video of a railway scene to be monitored, extracting a series of to-be-monitored original images from a monitoring video, processing the to-be-monitored original images through an existing Gaussian mixture model, and sequentially performing morphological processing, threshold adaptation and non-maximum suppression algorithm processing on output binary images to obtain an image area containing foreign matter invasion; and inputting the image area containing foreignmatter invasion into the YOLOv3, wherein the YOLOv3 outputs the foreign matter type and the foreign matter positioning position information in the corresponding original image according to the trained model weight. According to the method, the false alarm rate caused by the Gaussian mixture algorithm is reduced, the detection effect of the YOLOv3 algorithm on a long-distance target or/and a smalltarget is improved, and railway foreign matter invasion can be effectively monitored.

The invention discloses a method for identifying vehicles from GPS (global positioning system) data. The method includes steps of preliminarily labeling GPS data points; acquiring candidate transfer stations, and in other words, adding potential transfer stations into sets of the candidate transfer stations; acquiring true transfer stations and sectioning tracks; extracting characteristic acceleration change rates ACR; extracting characteristic time slice types TS and dividing user working peak hours into two types T_busy and T_idle; extracting 85% division speed and acceleration characteristics of road sections; extracting characteristics from the various road sections, inputting the characteristics into classification model random forest classifiers and outputting transport modes to obtain prediction results of the transport modes. The transport modes are output results. The method has the advantages that the effective characteristics are extracted according to laws presented by existing traffic, and the vehicle identification rate further can be increased by the aid of post-processing algorithms on the basis of ideology of track integrity.

The invention discloses an optimization method of a parallel pooling layer for optimizing the surface abrasion detection model of a pantograph carbon contact strip. The method comprises the followingsteps: 1) collecting a defect image on the surface of the pantograph carbon contact strip, and carrying out image preprocessing to obtain a data set; wherein the data set comprises training data and test data; the training data comprise two types of data, namely label data and label-free data; 2) building a semi-supervised convolution neural network under a CAFFE framework, and training the modelby utilizing the label-free data; 3) replacing an original pooling layer by a parallel pooling layer on the basis of a random pooling principle, subjecting the label data and the label-free data to differential sampling, and optimizing the surface abrasion detection model of the pantograph carbon slide plate on the basis of the semi-supervised convolution neural network. According to the invention, differential sampling is carried out on different attribute data by adopting the parallel pooling layer. The utilization efficiency of the semi-supervised convolution neural network on the label-free data is improved. The optimization effect of the surface abrasion detection model of the pantograph carbon contact strip is improved.

The invention belongs to the field of nondestructive testing, and discloses a manufacturing method of a multipurpose nondestructive testing natural defect test block, which comprises the following steps: forming a groove on a base material with the thickness of more than 12mm, and carrying out backing welding by adopting argon tungsten-arc welding, wherein both the base material and a weld joint can be magnetized; filling the grooves of the base metal through shielded metal arc welding to complete weld joint filling; in the filling welding process, making an incomplete fusion defect, an incomplete penetration defect, an air hole defect, a slag inclusion defect and a tungsten inclusion defect in the weld joint; determining a designed crack area on the weld joint, coating ferrous sulfide powder on the designed crack area, and then performing shallow penetration and remelting on the weld joint in the designed crack area; polishing the remelting area of the weld joint, and polishing off unmelted ferrous sulfide; and performing cosmetic welding on the surface of the weld joint to obtain the multipurpose nondestructive testing natural defect test block. The manufacturing process is simple, the cost is low, cracks with various defects and that are close to the natural forms are formed, and the training and teaching requirements are met.

The invention discloses an all-directional invasion detector which comprises an invasion detector body, infrared LED lamps, a connecting frame and a supporting column, wherein two infrared detection windows and a working indication lamp are arranged in the invasion detector body; the infrared detection windows are connected with each other through a connecting belt; a working indication lamp and the infrared LED lamps are arranged on the connecting belt; the invasion detector body is connected with the supporting column through the connecting frame; both the infrared detection windows are semispherical; the invasion detector body, the connecting frame and the supporting column are rotationally connected. According to the all-directional invasion detector, due to the infrared LED lamps, the invasion detector still has a good detection effect even in very dark light; the infrared detection windows form a spherical shape and the invasion detector body can rotate, so that the detector can be suitable for key cultural relics or warehouse protection with a relatively wide invasion detection range.

The invention discloses a semantic edge detection method based on deep learning. The core is a semantic edge detection deep neural network model based on multi-level feature fusion. The method comprises the steps of multi-level feature learning, multi-level feature extraction and multi-level feature fusion. The training comprises three steps of initializing model parameters, preparing a target data set and training an overall model. Compared with the prior art, the method has the following advantages: 1) the low-level features are gradually fused into the high-level features from bottom to top, the semantic discrimination capability of the fused high-level features is enhanced, and the missing underlying detail information is supplemented, so that the semantic edge detection effect is improved (2) training a multi-level feature learning module on the big data set in advance to enable the big data set to learn richer multi-level features on a target data set, and (3) expanding the dataset by adopting a data enhancement technology and preprocessing true value tagse to enhance the learning ability of the whole model.

The invention provides an ultrasonic probe correction method and system. The method comprises steps: the flight time of ultrasonic signals transmitted by all array elements in the ultrasonic probe under a set distance or set time is measured, a reference array element is selected, and a flight time difference/ratio/product between each array element and the reference array element is calculated; and according to the flight time difference/ratio/product, compensation parameters of each array element are calculated, and according to the compensation parameters, transmitting or receiving of the ultrasonic signals of each array element is compensated. According to the method and the system, compensation and correction on process errors of the ultrasonic probe array can be realized, and a basisis provided for compensation and calculation on beam transmitting and receiving by the ultrasonic system.

The invention belongs to the technical field of image recognition and detection, and particularly provides a tire X-ray defect detection method based on twin distance comparison, which comprises the following steps: acquiring a set of pictures of a standard picture, a normal picture and a defect picture of a specific tire pattern; labeling the sample set of the X-ray detection picture according todifferent types of pictures, and dividing samples with labeling information into a training set, a verification set and a test set; preprocessing the X-ray detection picture; establishing a Siamese Network model; training the Siamese Network model based on the training set, iterating the training process for multiple periods, and finally selecting parameters enabling the Siamese Network model tobe highest in accuracy; and after the obtained Siamese Network model is loaded, judging whether a defective part exists in the X-ray detection picture or not. Compared with a traditional method, the detection method has the advantages that the detection effect on unobvious defects can be greatly improved, the detection method is suitable for X-ray films shot by different models, tire patterns andX-ray machines, the burden of tire quality inspection personnel can be reduced, and tire quality control is facilitated.

The invention relates to an adaptive threshold value color image edge detection method based on an FPGA (Field Programmable Gate Array) and Kirsch. The method comprises the following steps that: firstly, converting collected image data into YCbCr, and carrying out Gaussian filtering and median filtering processing on a Y component in the YCbCr; and then, carrying out edge detection, carrying out morphological processing on an image subjected to the edge detection, synthetizing Y'Cb'Cr', and converting the Y'Cb'Cr' into an RGB888 format to be displayed in virtue of a VGA (Video Graphics Array). A Kirsch operator is taken as the reference of the edge detection, the edge detection of the adaptive threshold value color image is realized in virtue of the FPGA platform, and in addition, visual display is carried out through the VGA in the RGB8888 format so as to improve a color image edge detection effect. By use of the method, the defects in the prior art that image processing instantaneity is insufficient can be made up, and the edge detection is more flexible so as to be favorable for improving the accuracy of object edge detection.

The invention discloses a multi-probe and single-probe integrated test system and a multi-probe and single-probe integrated test method, which aim at solving the disadvantage of the existing antenna measurement system that measurement from a meter wave level frequency band to a millimeter wave level frequency band cannot be realized. The multi-probe and single-probe integrated test system is characterized by comprising a rotatable bearing platform, a multi-probe array device and a single-probe detecting device, wherein the multi-probe array device is positioned on one side of the bearing platform, and comprises a plurality of first probes which are distributed in a ring array and used for realizing frequency band detection between meter wave and millimeter wave levels; and the single-probedetecting device is positioned on one side opposite to the multi-probe array device, and comprises a second probe which can rotate around a tested object to form a detection arc surface and is used for detecting up-down extension of the meter wave level frequency band and the millimeter wave level frequency band. The multi-probe and single-probe integrated test system and the multi-probe and single-probe integrated test method realize the measurement of radiation feature data of the tested object from the meter wave level frequency band to the millimeter wave level frequency band, and are more convenient in measurement process.

The invention provides a method for detecting unexpected hot topics in Chinese microblogs. The method comprises the following steps: acquiring microblog posts of microblog sites through data interfaces of microblog sites, wherein the microblog posts comprise extracted texts and publish time; establishing a reference microblog post set according to the microblog posts; determining the microblog post set to be detected; adopting the binary coverage method for word segmentation; determining unexpected words, and establishing an unexpected word cluster; determining and ordering hot topics. The method can detect unexpected topics in massive Chinese microblog posts, and adopts the unexpected word and unexpected word cluster detection combined manner by utilizing the features of the unexpected topics in the microblog posts to accurately detect the unexpected topics.