1121results about How to "Improve detection ability" patented technology

The invention relates to pipeline defect ultrasound guided wave time reversal detection device and method. The method includes the following steps: selecting detecting frequency according to the detected pipeline corresponding free hollow column structure group velocity dispersion curve; inputting the frequency into the arbitrary function generator to generate center frequency used as single sound signal; sending the signal to each passage of exciting / receiving set to transducer unit; exciting longitudinal axis symmetry guided wave modal; sending the reflected signal to the computer and gaining reversal excitation signal by time reversal; repeatedly exciting guided signal to detect. The invention realize space and time focus for guided wave detection, greatly improve detection capability for little defect.

The invention relates to a radar detection technology, and provides a method and a device for detecting radar acquiring low-altitude wind profiles through combination of a frequency-modulated continuous-wave radar technology and a wind profile radar technology against the poor low-altitude detection capability of the existing equipment. One phased-array antenna is adopted to transmit serrated linear frequency-modulated continuous waves, and the other phased-array antenna is adopted to receive echo signals, thus forming a dual-antenna working mode of continuous-wave radar. The array surfaces of the two antennas are horizontally placed side by side with opening surfaces facing upward. Time-sharing scanning is carried out to form five detection wave beams which are the same with wind profile radar, and detection signals and data of the five wave beams are processed in the same way that signals and data of wind profile radar are processed so as to obtain a low-altitude wind profile of which the smallest height can be as small as 50m to 10m and other basic data products. A low-altitude continuous-wave wind measuring radar manufactured by adopting the method and the device of the invention has a wide range of applications.

This invention relates to one portable X wave section Doppler gas radar signal process method, wherein, the receive sub system processes wide band receive method on signals to form main path amplifier of low noise sound effect tube to form receive system together with reference signal amplifier ad lock phase vibration and stable lock ; adopting wide band for receive within range of 45MHz-75MHz automatically catching and tracing aim signals; Main path amplifier is composed of low noise sound effect, mixture device for selecting echo signal to output middle frequency signal; the reference signal amplifier is composed of mixture device and middle frequency amplifier.

The invention belongs to the technical field of microscopic spectrum imaging, and relates to a spectroscopic pupil laser confocal Raman spectrum testing method and device, wherein a confocal microscopic technology and a Raman spectrum detecting technology are combined. A spectroscopic pupil confocal microscopic imaging system is constructed by using rayleigh scattering light discarded in confocal Raman spectrum detection, high-resolution imaging and detection of a three-dimensional geometric position of a sample are realized; and a spectrum detection system is controlled by using an extreme point of the spectroscopic pupil confocal microscopic imaging system to be capable of accurately capturing Raman spectrum information excited by a focusing point of an objective lens, and further spectroscopic pupil confocal Raman spectrum high-space-resolution imaging and detection of image and spectrum integration are realized. The spectroscopic pupil laser confocal Raman spectrum testing method and device provide a new technical approach for high-space-resolution detection of the three-dimensional geometrical position and spectrum in a microcell, can be widely applied to the fields such as physics, chemistry, biomedicine, material science, environmental sciences, petrochemical engineering, geology, medicines, foods, criminal investigation and jewelry verification, and are capable of carrying out nondestructive identification and deep spectrum analysis of a sample.

The invention discloses an in-orbit moving target detecting method with space-based radar and infrared data fusion. The method includes the steps that when an infrared detecting system is used for detection, background suppression is carried out on each frame in an infrared image sequence by adopting a filtering method based on the morphology, self-adaptive threshold segmentation is carried out on images obtained after background suppression, a single frame detection result is extracted, a multi-frame target is determined through a tracking algorithm, and whether the detection result is a real target or not is judged to obtain the detection result of the multi-frame target; during radar detection, pulse doppler (PD) is adopted for processing. After time calibration and space calibration are carried out on the obtained infrared and radar target information, data fusion is carried out on the information by adopting a measurement fusion method, and track prediction is carried out on fused data to obtain the estimated position of the target.

The invention discloses a 360-degree dead-angle-free obstacle intelligent detection and early warning method for a vehicle, comprising the following steps that system errors are corrected through the external contour modeling of the vehicle and the correction of a sensor; surrounding areas surrounding the vehicle are classified into an effective scanning area, an ineffective scanning area and a dead angle area; and a trip computer acquires obstacle information through the vehicle-mounted radar sensor, calculates and obtains relative data of obstacles and the vehicle and running data of the vehicle relative to a stationary ground and outputs a two-dimensional top view plane of a vehicle roof from top to bottom. A driver can comprehensively know obstacles, vehicles and other objects surrounding the vehicle at any time through a display. The invention provides accurate environmental information for narrow road driving, lane switching driving and reversing of the vehicle and is a driving safety auxiliary method with extremely high intelligence and applicability.

The embodiment of the invention discloses a fraudulent trading detection method and a server. The method and the server comprise the steps that the server receives a message of a transaction request delivered through the terminal, and then according to the transaction data of a current request transaction and each input variable of a scoring model of the fraudulent trading, the corresponding value of each input variable of scoring model of the fraudulent trading can be confirmed and further the score value of the current transaction can be received. If the score value of the current transaction is bigger than the threshold value of the score value, the current request transaction can be confirmed as a fraudulent trading. The server can construct a scoring model of the fraudulent trading according to the transaction data of multiple historical transactions, the card files and the merchants files. Therefore, when the server gives scores to the current request transaction, the server can confirm the score value of current request transaction according to the data in the merchants files and card files, and judge if the current request transaction is a fraudulent trading. According to the fraudulent trading detection method and the server, the problems of detection leaking in the bank card fraudulent trading can be efficiently avoided and the detection effect of the fraudulent trading can be improved.

The invention relates to a plate-structure ultrasonic array imaging method based on the virtual focusing, belonging to field the non-destructive detection. The method comprises the step of processing receiving data of a transducer array to realize default pointing and focusing detection in any directions within the range of 360 degrees around the transducer, so that the problem of large-scale detection can be solved. When a wave beam of signals points and deviates to a direction in which a default position exists, all signals are focused at the default position, the energy of reflecting signals caused by the default is the strongest in the direction, thus the direction information of the default is obtained, and the signal-to-noise ratio is improved. Each sensor unit in a sensor array is connected with an independent and multi-channel excitation / reception conversion switch, so that the excitation and the reception of test signals are very flexible.

The invention relates to passive radars, electric wave transmission and wireless communication, provides a broadband recognition passive radar facing a complicated heterogeneous wireless communication irradiation source in the future, and is suitable for the development tendency of a new method for searching by a new theory in the field of the passive radars at the present.. According to the technical scheme, a broadband recognition passive radar system architecture design method comprises the following steps of: taking a complicated heterogeneous wireless communication emission signal as a passive radar irradiation source; constructing a passive radar irradiation source library based on a future broadband wireless communication technology through machine learning; then supplying a multi-system and multi-standard compatible broadband recognition passive radar system architecture; and finally, dynamically selecting an optimal waveform as the passive radar irradiation source according to an environment feedback and prior information supplied by the irradiation source library, and realizing heterogeneous fusion and preferential work of a complicated heterogeneous irradiation source. The method is mainly applied to design and manufacturing of the passive radars.

The invention belongs to the technical fields of optical microscopy imaging and optical precision measurement and relates to a laser differential confocal spectrum microscopy tomography device which mainly comprises a Raman spectrum analysis part (25) and an objective lens (6), a polarizing spectroscope (7), a one-quarter glass slide (8) and a measurement objective lens (9) which are sequentially arranged along the optical path; the laser differential confocal spectrum microscopy tomography device further comprises a differential confocal detection part (26) which is positioned in the reverse direction of the reflection direction of the polarizing beam splitter. The differential confocal detection part is used for measuring the geometric position of a micro-area of a sample and focusing the sample to obtain image information of the micro-area of the sample; the Raman spectrum analysis part is used for analyzing the material spectrum of the detected area of the sample to obtain component information of the micro-area of the sample; the combination of the two parts can realize the nano-level micro-area spectrum measurement of the sample and simultaneously obtain the geometric feature and the material component information of the micro-area of the sample. The laser differential confocal spectrum microscopy tomography device provides a powerful observation means for bio-medicine, material science, high-energy physics and other forefront subjects.

The invention relates to a light-splitting pupil laser differential motion confocal Brillouin-Raman spectrum measurement method and device and belongs to the technical field of microscopic spectrum imaging. According to the method and the device, the high-resolution imaging of a three-dimensional geometric position of a sample is realized by constructing a light-splitting pupil confocal microscopic imaging system through using rayleigh scattering light abandoned in the confocal Raman spectrum detection; a zero crossing point of a light-splitting pupil laser differential motion confocal microscopic imaging device and a focal point of the zero crossing point are used for corresponding to the property to control a spectrum detector to accurately capture Raman spectrum information excited by a focus point of an objective lens so as to further realize the high-precision detection, namely image and spectrum combined high-space resolution detection on a microcell geometric position and spectrum information of a sample; meanwhile, the resolution capability and a measuring range are considered effectively; the characteristics of complementary advantages of a confocal Raman spectrum detection technology and a confocal Brillouin spectrum detection technology are utilized to design a confocal spectrum detection scheme for detecting by adopting a Raman spectrum and a Brillouin spectrum so as to realize the comprehensive measurement and de-coupling of multi-performance parameters of materials.

The invention discloses a near space moving target positioning method based on an unmanned aerial vehicle, which belongs to the technical field of unmanned aerial vehicle target positioning. The method comprises the following steps: firstly, an infrared target detector loaded on the unmanned aerial vehicle is used for acquiring a target angle measurement value, in combination of an unmanned aerial vehicle navigation system, the position and the attitude information of the unmanned aerial vehicle are outputted, and the angle value of the connection line between the target and the unmanned aerial vehicle in an earth-rectangular coordinate system is calculated; then, according to the calculated angle value, the least-squares solution of the target in the earth-rectangular coordinate system is solved; and finally, the solved rectangular coordinates are converted to the longitude, the latitude and the height. The method can realize high-precision positioning of the near space target when a laser range finder or a radar can not be used normally, and can be applied to multi-unmanned aerial vehicle formation cooperative positioning.

The invention belongs to the technical field of microscopicspectral imaging detection, and relates to a split pupil laser differential confocal Raman spectrum test method and device. According to the test method and device, a split pupil laser differential confocal microtechnique and a laser Raman spectrum detection technique are organically combined, precise imaging of three-dimensional geometrical positions is realized through segmentation focal spot differential detection, the optical path structure of a traditional differential confocal microscopic system is simplified, advantages of an original laser differential confocal system and a split pupil confocal system are inherited, and multi-mode switching and processing of split pupil laser differential confocal microscopic detection, laser confocal Raman spectrum detection and laser differential confocal Raman spectrum detection can be realized only through softwareswitching processing. The test method and device provide a new technological approach for detection ofnanoscale microcell three-dimensional geometrical positions and spectrum, can be applied to fields of biomedicine, industrial precision detection and the like, and has the broad application prospect.

The invention provides an LFMCW radar detection moving human body target method based on a human body model. The method comprises the steps that analog-to-digital conversion is carried out on an LFMCW radar echo beat signal; FFT processing is carried out on each PRI echo beat datum to acquire the LFMCW radar echo beat signal frequency domain data when a human body slowly moves, and rearranging is carried out in a slow time-distance domain two-dimensional plane; a Boulic human body walking model is constructed; according to the human body moving gait frequency and an initial phase range, a number of human body trunk moving models with different parameters are constructed on a gait frequency-initial phase two-dimensional plane; an LFMCW radar echo nonlinear phase compensation signal is reconstructed, and phase compensation is carried out on the rearranged data; FFT processing is carried out to acquire a four-dimensional space processing result; whether a target is in a distance unit is judged through constant false-alarm detection processing; and the information of distance, velocity and gait frequency of the target are acquired in the presence of the target. According to the invention, the signal-to-noise ratio of LFMCW radar detection is improved, and the human body target detection performance in strong ground clutter is improved.

The present invention discloses an eddy current detecting sensor and a detecting method based three giant magneto-resistances. An excitation coil is adopted to generate an alternating magnetic field and arranged above a detected object. A giant magneto-resistance I, a giant magneto-resistance II and a giant magneto-resistance III with pair-wise orthogonal magnetic sensitivity directions between the excitation coil and the detected object. Wherein, the giant magneto-resistance I is connected with an amplifying circuit I. The giant magneto-resistance II is connected with an amplifying circuit II. The giant magneto-resistance III is connected with an amplifying circuit III. The amplifying circuit I, the amplifying circuit II and the amplifying circuit III output voltage signals corresponding to the giant magneto-resistance I, giant magneto-resistance II and giant magneto-resistance III. The voltage signal comprises eddy current distribution information of the detected object. The eddy current detecting sensor based on three giant magneto-resistances can not only bring high sensitivity and spatial resolution to deep or multilayer flaw detection but also benefit quantitative defect analysis.

The invention relates to a dual polarization radar arbitrary polarization wave on-line calibrating method. A polarization calibrating channel is constituted by a digital calibrating unit, a digital-to-analog converter, a frequency converter, a calibrating horn, a main feed source horn, and a receiving channel. The digital calibrating unit is substantially an arbitrary waveform generator, and is used to generate standard continuous wave sinusoidal signals, noise signals, arbitrary digital waveforms, and delay of emission signal transcripts. When a polarization radar works in a calibrating mode, an output end of an emitter is used to couple a few of signals, which are used as transcripts of the emission signals. The signals are converted into digital signals after down conversion operation, filtering operation, and ADC conversion operation, and are used as digitalized emission sampling signals, which are input in the digital calibrating unit. After the amplitude conversion processing, the phase conversion processing, the frequency spectrum conversion processing, the delay conversion processing, and other conversion processing of the signals in a numerical field, the signals are converted into analog signals, and the analog signals are converted into radio frequency signals by up conversion operation, and then are irradiated on a main feed source by the calibrating horn.

The invention provides a method for measuring large-area soil humidity, which comprises the following step of: mounting a vertical polarized receiving antenna, a GPS (Global Positioning System) receiver and a computer for carrying out signal processing. The measuring process is carried out by utilizing the interference effect of a GPS signal and a reflecting signal thereof. By utilizing the widely existing GPS signal as a signal source, the method can be used for carrying out all-weather large-area measurement, has low cost and more further simplifies a receiving device by utilizing the interference effect.

The invention discloses a river way multi-sensor fusion upstream and downstream pollution early warning system and method. A river way part includes a water quality on-line monitoring buoy nodes deployed on multiple parts of the upstream and the downstream of a river way, and a shore part includes a wireless gateway and a cloud server. The water quality on-line monitoring buoy nodes are used for real-time collection, storage and transmission of water quality data; the system includes a water quality multi-sensor system, a water flow sensor, a real-time clock module, a microprocessor, a wireless communication module, an antenna, a storage module and a power source module. A river way multi-sensor data fusion analysis model is established by the cloud server, the multi-sensor data fusion method is used for analyzing whether the pollution exists in the river way, classifying pollutants and estimating the location of a pollution source, the upstream and downstream relations are combined for on-line monitoring and classified early warning on the river way water quality, whether the river way is invaded by the pollutants is judged, and the location of the pollution source is estimated. The system and the method perform on-line monitoring by multiple water quality parameters, and have the advantages of low energy consumption, high detection rate of the pollutants and wide applicationrange.

The invention relates to an infrared thermal imager and a method for detecting an infrared objective image by using the same. The infrared thermal imager comprises an infrared optical system, an imaging circuit component, a driving circuit and a monitor, wherein the infrared optical system comprises a varifocal objective; and a refrigerator projection screen, a combination optical mirror and an infrared detector are sequentially arranged behind the varifocal objective. The method for detecting the infrared objective image by using the infrared thermal imager comprises the following steps of: focusing an image of a scene; scanning the image; focusing, and converting an electric signal; and correcting temperature in real time, and controlling. The invention has the advantages that: the infrared thermal imager is smart in structure, simple and convenient to operate, quite high in reliability and accurate in temperature control; the problem of temperature difference between a reference calibration source and the scene in the prior art is solved; effects of real-time performance and self-adaptation in correction of the current scene are good; the uniformity of the corrected image is guaranteed; a correction effect is greatly improved; and the quality of the image is high.

The invention discloses a non-contact static detection device, which comprises a static sensor, an electronic switch and an electric signal processing unit, wherein the static sensor comprises a sensing electrode and a coupling electrode; the sensing electrode is a metal static signal receiving surface; the coupling electrode is a metal electrode; the sensing electrode and the coupling electrode construct a capacitor of fixed capacitance; the sensing electrode is electrically connected with the electronic switch; and the electric signal processing unit is used for grounding or suspending the sensing electrode alternatively by controlling the electronic switch, and is also used for performing subsequent processing on an alternating signal acquired from the coupling electrode to convert the processed alternating signal into an electric field value. The non-contact static detection device has the characteristics of no mechanical part, high coupling capacitance variation quantity, high sensitivity, flexible sensor design and easiness in realizing; and by selecting the static sensor, static electricity or weak static electricity in tiny regions can be measured in real time and absolutely.

The invention provides a self-driven photodetector based on an FTO / ZnO nanorod / CH3NH3PbI3 / MoO3 / Au structure and a preparation method thereof. The specific structure comprises an FTO layer and a ZnO nanorod which is an electron transporting layer and a hole blocking layer. CH3NH3PbI3 is a perovskite light absorbing layer. Semiconductor oxide MoO3 is a hole transporting layer and an electron blocking layer. A metal electrode is made of an Au film. The self-driven photodetector is synthesized through spin coating and water bath and is prepared through evaporation and the like. According to the invention, an organic-inorganic hybrid heterojunction structure formed by the ZnO nanorod / CH3NH3PbI3 is used; the semiconductor oxide MoO3 is the hole transporting layer; the self-driven photodetector has the advantages of high response degree and sensitivity; the response rate and the detection rate are respectively up to 24.3A / W and 3.56*1014cmHz1 / 2 / W; the self-driven photodetector has a certain degree of self-driven capacity and does not need the drive of external bias, which is conducive to energy conservation; the performances are far more than the performances of a currently reported Si-based detector; near-ultraviolet and visible infrared detection can be realized; and the self-driven photodetector has the advantages of simple operation steps, low experiment cost and good application prospect.

The invention relates to an orientation identification method of ultrasonic array crack defects based on vector full-focusing imaging, and belongs to the field of nondestructive testing. According to the method, vector full-focusing treatment is performed on signals received by a transducer array so as to obtain a global vector diagram, the positions of the defects are determined by the strongest energy of reflected signals at the defects, then local vector diagrams of the defects are extracted according to the positions of the defects, and finally the directions of the vectors are determined by utilizing the directions of vectors in the local vector diagrams. According to the principle of vector full-focusing imaging, a plurality of sub-arrays are constructed in the transducer array, the unit direction vector at any imaging point is calculated through the sub-arrays, and echo amplitude after compensation at any imaging point to the transducer array is vectorized by utilizing the unit direction vector, so that amplitude vector at any imaging point is obtained.

The invention discloses a crossed combinational dual-period grating for quantum well infrared detector. By arranging two sizes of gates across, the invention responses to two wave bands and acquires the information of the two wave bands. The cross array makes a very good response uniformity of the new-type gates on the whole surface, and the two types of gates are distributed as the Chinese character 'pin', which makes the photolithography linewidth up to the maximum and thereby enhances the precision of photolithography patterns. The square gate hole is used to prevent astigmatism; the optimized parameter design enhances the coupling efficiency of the gates. The invention is applied to dual-color quantum well infrared detectors, is able to acquire the objective information more accurately in two wave bands simultaneously, more effectively inhibits the complicate background, and then enhances the detecting effect of objectives. The invention has the advantages of simple fabrication, low cost, and easy realization of mass and serial production.

The embodiment of the invention relates to the technical field of data processing, in particular to a transaction monitoring method and a transaction monitoring device, which are used for improving the recognition capability of abnormal transaction behaviors. The method provided by the embodiment of the invention comprises the steps of obtaining monitoring transaction flow in a monitoring time period; forming a transaction directed graph by utilizing the monitoring transaction flow, determining suspicious transaction communities from the transaction directed graph, and enabling one suspicioustransaction community to comprise a plurality of suspicious transaction subjects; and for any suspicious transaction subject, calculating an abnormal risk value of the suspicious transaction subject by using an abnormal transaction algorithm model according to the monitoring transaction flow of the suspicious transaction subject, and performing clustering on the suspicious transaction subject by using the abnormal transaction algorithm model to obtain a training sample set which is similar to the training sample set which is obtained by not depending on labels.

The invention discloses a rare earth oxide solid solution ceramic scintillator and a preparation method thereof. The main component of the ceramic scintillator is Gd2xLu2yY2(1-x-y-z)Eu2zO3 (the x is more than or equal to 0.1 and less than or equal to 0.6, the y is more than or equal to 0.1 and is less than or equal to 0.4, and the z is more than or equal to 0.01 and less than or equal to 0.1), and the ceramic scintillator has a crystal structure with a cubic Ia3 point group. Ceramic powder can be synthesized by a chemical coprecipitation method. The coprecipitation method adopts ammonia, ammonium hydrogen carbonate or a mixed solution of the ammonium water and the ammonium hydrogen carbonate as a precipitating agent, titrates the precipitating agent into a solution of gadolinium nitrate, lutecium nitrate, yttrium nitrate and europium nitrate to obtain a precipitate, and then the precipitate is dried and calcined to obtain nano-powder. The obtained powder is pressed into a ceramic blank through an isostatic compaction method, then the pressureless sintering is performed in vacuum or hydrogen atmosphere, the sintering temperature is between 1,600 and 1,900 DEG C, transparent Gd2xLu2yY2(1-x-y-z)Eu2zO3 ceramic can be obtained, a ceramic product with the needed dimension is prepared after the cutting, grinding and polishing, and the transmission rate of the ceramic in a visible region (400-800 nanometers) is more than or equal to 65 percent. The ceramic scintillator emits red light with a main wavelength of 610 nanometers under the excitation of ultraviolet light or X rays, and can be used for scintillating materials of imaging and detection of medical and industrial X rays.

The invention relates to an underground construction advanced detection method and system based on resistivity of multiple same source electrode arrays. The underground construction advanced detection method and system is suitable for detection of unfavorable geology in a borehole-blasting method of underground construction including tunnels or underground tunnels or the construction process of a tunnel boring machine (TBM). The underground construction advanced detection method comprises the steps of (1) constructing an array type measurement electrode system on a working surface; (2) forming a power supply electrode system on a side wall, and arranging the other power supply electrode B and measurement electrodes N to the position far away behind the working surface; (3) providing positive and negative current through a power supply electrode A positioned on the working surface and the power supply electrode B positioned far away behind the working surface, utilizing each of measurement electrodes M in the array type measurement electrode system to measure electric potential respectively, and obtaining the electric potential difference of the measurement electrodes M and the measurement electrodes N in the working surface and fracture surfaces, wherein the step is repeated continuously until the power supply electrode system moves to the preset position, and the detection is completed; (4) performing inversion on the measured electric potential difference data by adopting least square linear inversion iteration process based on smooth constraint to obtain three-dimensional resistivity images of a geologic body in front of the working surface.

The invention belongs to the technical field of non-destructive testing and relates to an ultrasonic testing method used for composite material foaming structures. According to the method, ultrasonic waves formed by a large number of shock pulse waves in the composite material foaming structures are utilized, and by constructing transfer functions of the ultrasonic waves in the composite material foaming structures, the arithmetical relationship between the ultrasonic waves and different part defects of the composite material foaming structures is set up to carry out defect discriminating and ultrasonic testing; the shock pulse ultrasonic waves of different types, intensities and frequencies and ultrasonic testing achieving methods can be selected, and the different composite foaming structures can be tested in an ultrasonic mode. Actual testing results show that the sensitivity, the resolution and the ability of composite material foaming structure testing are remarkably improved, the defects and the combination conditions of skin, skin / foam cementing areas, foam / foam cementing areas and foam areas can be tested accurately, the testing reliability is greatly improved, and the foam testing thickness can be 100 mm.

The present invention belongs to the field of gas detecting technology. The methane concentration detecting method includes using light source in certain spectrum range and resonance cavity with two high reflectivity inner surfaces; detecting the light from the light source and through the detected gas and the resonance cavity; regulating the length of the resonance cavity to detect the maximum light strength and the minimum light strength; and calculating the methane concentration with the formula. The detecting device includes LED light source, collimating lens, converging lens, Fabry-Perot etalon, detection system, computer, and signal generator. The present invention has low cost, convenient optical path collimation and high sensitivity.

The invention provides a self-adaptive weight deep learning target classification method based on feature fusion. The method comprises performing target coarse detection; extracting image convolutionfeatures and HOG features, and performing dimension expansion processing on the HOG features; embedding the SENet into a Resnet network framework, and establishing a network framework for extracting multi-feature weights of the image; calculating adaptive weight vectors of the convolutional features and the HOG features, making a feature fusion strategy, and calculating image fusion features; andestablishing a multi-target classification framework based on the precise binary classification network set. According to the method, image convolution features and HOG features are fused, adaptive weight vectors of the image features are extracted, deep learning network configurations and parameters are designed, an accurate classification network is constructed, the network obtains more candidate frames by reducing a score threshold value, and the recall rate of target detection is increased; by designing a plurality of binary classification networks, the method has higher accuracy in termsof multi-classification problems.

The invention discloses a ferroelectric local field enhanced two-dimensional semiconductor photoelectric detector and a preparation method. The detector is characterized in that: the structure of the detector from top to bottom comprises a substrate, a two-dimensional semiconductor, a metal source and drain electrodes, a ferroelectric functional layer and a semi-transparent metal upper electrode in sequence. The preparation method of the detector comprises the steps: a transition-metal chalcogenide two-dimensional semiconductor is prepared on the substrate, an UV lithography or an electron-beam lithography is adopted and combines stripping technology to prepare a metal electrode serving as a source electrode and a drain electrode of a semiconductor channel, a ferroelectric thin film is prepared on the structure, a semi-transparent or transparent electrode is prepared on the ferroelectric thin film, so that a two-dimensional semiconductor detector structure is formed. The detector can make the two-dimensional semiconductor channel background carriers to be completely exhausted through using polarized ferroelectric materials, tiny voltage is applied between the source electrode and the drain electrode, and photoelectric detection is realized through detecting the current signal change under light. The detector has advantages of high sensitivity, fast response, good stability, low power consumption, wide spectrum detection and the like.