82results about How to "Reduce missed judgments" patented technology

An automatic snapshot system for highway traffic offence comprises a front snapshot unit, a network transmission unit, a detection unit and a management server, wherein the front snapshot unit obtains real-time video in a detected scene, the network transmission unit transmits the real-time video to the detection unit; the detection unit after obtaining the video uses object segmentation and trace tracking and analysis to find vehicles having abnormal behaviors such as reversing, going in a direction not allowed by traffic regulations and illegal parking, providing the position of the target vehicle, controlling a holder to rotate so that the license plate of the target vehicle is in the image capture center of the front snapshot unit, controlling the front snapshot unit to zoom a lens and adjust the size the target in the image to facilitate license plate identification and analysis; and after the license plate is determined, the front snapshot unit controlling an electric zoom lens to restore to a wide angle monitoring mode and snapshot distant view, close view and license plates, and uploading the snapshot results to the management server unit.

The invention discloses an automatic snapshot device and method for illegal parking. The automatic snapshot device comprises a front-end snapshot unit, a transmission network and a management server. The front-end snapshot unit comprises an electric zoom lens, an intelligent camera and a tripod head. The electric zoom lens is arranged at the front end of the intelligent camera and connected with the intelligent camera. The intelligent camera is arranged on the tripod head and further connected with the management server through the transmission network. The automatic snapshot method comprises the steps that the front-end snapshot unit monitors the whole monitoring area through a wide-angle lens and tracks the motion trail of all vehicles within the monitoring area; when a parking event is found through the trail tracking mode, the barycenter and area of the parking vehicle are determined; the rotation angle and pitch angle of the tripod head are determined through geometric operations; the tripod head is controlled so that the target can be located in the center of an image; the size, in the image, of the target is adjusted through lens zoom control so as to be suitable for image analysis, namely license number recognition; after the license number recognition is completed, the intelligent camera controls the electric zoom lens to restore to a wide-angle monitoring mode.

The invention provides a vehicle illegal parking detection method based on a convolutional neural network. The method comprises the steps that (1) a no-parking region in a video frame image, a presetviolation early-warning threshold value and a preset violation threshold value are set respectively; (2) target vehicles in the set no-parking region are detected through a convolutional neural network model, and vehicle information is recorded; (3) whether a current vehicle set A in the step (2) and a historical vehicle set B contain vehicles which can be matched is judged; (4) whether delay timeof a current target vehicle Ai is greater than the preset violation early-warning threshold value in the step (1) is judged; (5) whether the early-warned delay time of the current target vehicle Ai is greater than the preset violation threshold value in the step (1) is judged; and (6) the vehicle information in the historical vehicle set B is updated, a next video frame image is input, the step (2) is entered to start vehicle violation detection on the next video frame image, and the operation from the step (2) to the step (5) is repeated.

The invention discloses a digital image noise reduction method and device. The noise reduction method includes: performing noise analysis on an input image, calculating the difference frame/field image between input image frames or between two fields separated by one field, and based on the difference frame Image analysis of the frame/field noise statistical characteristics; perform noise analysis, calculate noise power, determine noise signs and determine noise classification; adaptively select frame/intra-field noise reduction or/and frame/inter-field noise reduction methods. The device for digital image noise reduction includes: frame memory, noise analysis unit and frame/field noise reduction unit. Since the present invention selects and adopts frame/intra-field and/or frame/inter-field noise reduction methods by analyzing the noise characteristics of difference frame/field images, and selects appropriate weights according to the characteristics of noise and image information, the noise reduction Finally, images with better visual effects are obtained.

The invention provides a forest smoke and fire detection algorithm based on video image analysis which comprises eight steps to detect the smoke and haze in a video: preprocessing the video images; performing background modeling; carrying out morphological treatment; segmenting the motions, tracking the regions, determining the regions continuously; extracting the characteristics and determining the smoke and haze. According to the method of the invention, targeted to the characteristics of the forest smoke and haze video images, the method makes improvement in the background modeling technology in currently available video image analysis method by using the random clustering to acquire the initial model from the neighboring domain of the pixel points and then, through some samples in the neighboring domain of the pixel points, the background points are updated. This overcomes the problem of misjudgment caused by the excessive concentration of the selected samples and further accelerates the efficiency for smoke and haze detection. Requiring a small amount of computation, the method can be used to directly determine the smoke and haze from the video acquisition end so as to increase the computation efficiency while ensuring the detection accuracy. Therefore, the detection cost can be reduced; better timeliness can be achieved; and therefore, the method can find wider applications.

The invention discloses a separating and identifying method for multiple human body objects in the distance direction of a UWB impact biological radar. Correlation coefficients of point signals in different distances in the fast time direction are solved on the basis of the characteristic that correlation of breathing signals of an object itself is high and correlation of breathing signals of different objects is low; and a correlation coefficient spectrum obtained by calculation is combined with certain signal pretreatment to separate and identify the human body objects in the distance direction. The method can improve the separating and identifying performances for multiple objects in the distance direction of the biological radar, judgment leakage is reduced, more accurate one-dimensional distance-direction object position information is provided for detecting and 2D positioning of objects distributed randomly, the detecting and positioning performances of the domestic multi-object biological radar are improved, and the method has great significance in approaching the international advanced rank.

The present invention provides a dirty data identification method and device, a data cleaning method and device, and a controller. The dirty data identification method comprises the following steps: extracting a domain rule base, wherein the domain rule base comprises one or a plurality of first decision rules and one or a plurality of second decision rules; identifying dirty data in a data fieldaccording to the domain rule filtering the data field by the first decision rule, and judging the data field as dirty data if the data field matches any one of the first decision rules; if the first decision rule is not matched, filtering the data field using the second decision rule. The dirty data identification method of the invention improves data cleaning efficiency and cleaning accuracy.

This invention discloses one method to correct measurement data clear error and random error, which comprises the following steps: using Bayes method to extracting measurement list reliability from historical data; then leading clear error test as base for test error; using error test to get information and to reduce correction computation volume. This invention uses meter historical operation data as reliable base to determine one batch of high reliable computation volume.

The invention discloses an S1-MME data-based method of determining mobile user rapidly and accurately, belonging to the technical field of data analysis. According to the method, a pre-determined useris screened out by collecting S1-MME interface data in an LTE mobile communication network, and a practical traffic route and a travel distance of the user are simulated and calculated by combining an algorithm calculating a projection of a base station cell set covering the traffic route on a geographic traffic route and a cell coverage track segmentation algorithm. The method replaces a mannerof calculating a user movement speed by using cell space in the prior art, so that the movement distance and speed are determined more accurately; and the user is determined by using both a maximum span determining method and a reverse slide window determining method, so that the determining accuracy is improved, manual intervention factors are reduced, and false negative and false positive problems resulting from the fact that fewer cells cover the traffic route or traffic routes of different mobile attributes are adjacent are solved.

The invention provides an abnormal flow alarm decision-making system. The system comprises: a traffic classification component, which performs classification statistics on received data messages in astatistical period according to each of multiple dimensions at least including the traffic of the data messages, the protocol proportion and the number of source IPs, and respectively stores classification statistics results in a memory; a scoring component which is used for attributing the classified and counted result of each dimension to different threshold ranges of the corresponding dimension, and endowing the received data message with scores under the corresponding dimension based on the attributed different threshold ranges; and an exception judgment component which accumulates the scores corresponding to all dimensions, and judges the traffic state of the data message received in the statistical period corresponding to the accumulated score exceeding the normal threshold score asan exceptional traffic state of different levels.

The method comprises: deciding if the performance of high speed data service can satisfy the need of coverage at the switch area; if yes, then deciding if the neighboring cell allows making the switch of high speed data service channel; if yes, then directly switching to the high speed data service channel of neighboring cell, and making traditional cell switch; otherwise, using the normal channel to make switch. The invention also reveals a switch unit comprising a decision unit used for deciding if the high speed data service can satisfy the need of coverage at the switch area; a switch processing unit used for directly switching to the high speed data service channel of neighboring cell, and making traditional cell switch; otherwise, using the normal channel to make switch or using normal channel to make switch.

The embodiment of the invention discloses a power grid intelligent monitoring alarm method and a device. The method comprises the following steps that: through a thermal imaging module by a first predetermined interval, a visible light imaging module carries out timed shooting on a monitored object with a second predetermined interval; the image recognition technology is used to analyze a shot image, and whether an abnormal condition exists or not is judged; and an alarm is carried out when the abnormal condition exists. Thus, through combining visible light imaging and thermal imaging, the 7*24 all-weather monitoring can be realized, through using the image recognition technology, the judgment and alarm of an abnormal condition can be automatically realized.

The invention discloses a method for diagnosing a stator inter-turn short circuit fault of a doubly-fed fan based on a D-S information fusion algorithm, and belongs to the technical field of fault detection of a power system doubly-fed fan. The diagnosing method comprises the following steps: analyzing a characteristic mechanism of the stator inter-turn short circuit fault, establishing a MATLAB/Simulink-based stator inter-turn short circuit fault model for the doubly-fed fan, determining fault characteristic frequencies of a rotor-side current and a rotor instantaneous power spectrum, completing amplitude estimation of the two fault characteristic frequencies by the Prony algorithm, and further performing D-S information fusion to determine the fault, which effectively improves the faultmode identification ability, and has a reference value for the fault research of large wind farms.

The invention provides an anti-dry-burning control method and device and a gas stove. The anti-dry-burning control method includes the steps that information relevant to current pot bottom temperatureis acquired; information relevant to current cookware weight is acquired; and according to the information relevant to the current pot bottom temperature and the information relevant to the current cookware weight, whether a dry burning condition is achieved currently or not is judged, and if yes, a gas valve is controlled to be closed. According to the anti-dry-burning control method, due to theadoption of the dual detection standards of the pot bottom temperature and the cookware weight, misjudgment and missing judgment on the dry burning condition in a pot can be effectively reduced.

The invention provides a similarity threshold determination method considering the average fluctuation ratio of voltage curves. The method comprises the following steps of: A, obtaining distribution transformer related data which comprise corresponding relations between a 10kV line and distribution transformers, the names of the distribution transformers, the address codes of the distribution transformers, and the outlet voltage data of the distribution transformers; B, performing dimension reduction processing on the outlet voltage data of the distribution transformers acquired in the step A;C, calculating the average fluctuation ratio of voltage curves according to the outlet voltage data of the distribution transformers subjected to the dimension reduction processing in the step B; D,calculating the similarity of the voltage curves of the transformers of a transformer area under the same 10kV line according to the outlet voltage data of the distribution transformers preprocessed in the step B; and E, grading the similarity thresholds of the voltage curves according to the calculation results in the step C and the step D. By means of the method, the similarity threshold valuesof the voltage curves can be graded, and then the accuracy of similarity judgment is improved.

The invention provides a method for detecting pollution of a contact surface of a mask and a mask table. The method comprises steps as follows: the mask and a wafer are provided, and the mask is placed on the mask table of a photoetching machine and aligned to the wafer; S2, reference values of alignment compensation parameter of the mask are collected; S3, the steps S1 and S2 are repeated, multiple reference values of the alignment compensation parameter are collected, and an upper limit value and a lower limit value of the alignment compensation parameter are set according to the multiple collected reference values of the alignment compensation parameter; S4, the step S1 is repeated, an alignment compensation parameter value of the alignment is collected, if the alignment compensation parameter value exceeds the upper limit value or the lower limit value, the photoetching machine stops working and the mask table or the mask is cleaned, and if the alignment compensation parameter value is between the upper limit value and the lower limit value, the photoetching machine works continuously. According to the method for detecting pollution of the contact surface of the mask and the mask table, the condition of the contact surface of the mask and the mask table can be monitored in real time.

The invention discloses a display panel, a driving method of the display panel and a display device. The display panel comprises a display area and a non-display area, the non-display area is providedwith at least two signal input ends, at least one control signal input end, at least two detection signal lines and detection circuits, and the display area is provided with a plurality of data lines; the control ends of the detection circuits are electrically connected with the control signal input ends in a one-to-one correspondence manner; the input end of each detection circuit is electrically connected with one signal input end through one detection signal line; the output end of each detection circuit is electrically connected with different data lines; an input signal provided by at least one signal input end provides a black picture driving signal for the display panel, and an input signal provided by at least one signal input end provides a bright picture driving signal for the display panel. The intuition of crack detection of the display panel can be improved, and the accuracy of crack detection of the display panel is improved.

The invention discloses an oil-immersed capacitor oil leakage detection device. The oil-immersed capacitor oil leakage detection device adopts an ultraviolet light source, a first laser frequency screening device, a laser beam splitter, a data acquisition device and a data analyzer, wherein the first laser frequency screening device is used for screening single-frequency ultraviolet light from theultraviolet light source; the laser beam splitter is used for converting single-frequency ultraviolet light into multiple beams of parallel single-frequency ultraviolet light; the data acquisition device is used for receiving ultraviolet light radiated by the capacitor to be detected; the data analyzer is used for receiving the digital signal, and converting the digital signal into image data; the first laser frequency screening device is arranged between the ultraviolet light source and the laser beam splitter; the to-be-detected capacitor is arranged between the laser beam splitter and thedata acquisition device; and the data analyzer is electrically connected with the data acquisition device. The oil-immersed type capacitor oil leakage detection device does not depend on manual inspection, can accurately distinguish stains and oil marks by means of photoelectric conversion, and reduces misjudgment and leakage judgment.

The invention relates to a rainfall prediction method based on an improved decision tree algorithm. The method comprises the steps that collecting meteorological data of various regions in several years and corresponding grade data; normalizing meteorological data to obtain a corresponding normalized data set, and dividing the normalized data set into a training set and a test set in proportion; substituting the training set into the improved decision tree network for training, substituting the test set into the trained decision tree model, checking a training result, finally inputting to-be-tested data into the trained decision tree network for prediction, outputting a result, and evaluating the rainfall grade of the to-be-tested data. An independent variable average influence value is adopted for selection by taking the decision tree as the core, selection is made according influence of attribute on the result, the attributed with high influence degree is selected for branching, theimproved decision tree algorithm is used for training, mass data is fully utilized, the prediction accuracy is improved, and misjudgment and judgment neglect are reduced.

The invention discloses an automatic appearance inspection apparatus for a chip inductor. The automatic appearance inspection apparatus comprises a base; one side of the base is provided with a fixed plate; a computer input keyboard is arranged on the fixed plate; the side, perpendicular to the fixed plate, of the base is provided with a support; the support is provided with a computer through a fixed part; the side, far away from the support, of the computer is provided with a supporting seat; the supporting seat is mounted on the base; the side, far away from the base, of the supporting seat is equipped with a feeding device; a rotating disk is arranged on the central part of the upper end surface of the base; the side, far away from the base, of rotating disk is provided with a stepping motor; the edge of one side of rotating disk is provided with a material recovering box; and the upper end of the material recovering box is provided with a discharging device. The apparatus provided by the invention employs a computer integrated and programmed design and is simple to operate; after connection with a networking wire, the apparatus can carry out remote monitoring and remote data acquisition and realize unmanned operation; and the apparatus is simple to operate, easy to produce and suitable for extensive promotion.

The invention discloses an automatic identification method and device for coast erosion and sedimentation based on shoreline data, and belongs to the field of ocean monitoring. The method comprises the following steps: firstly, acquiring coastline vector data of two time phases, and respectively performing topology check on the coastline vector data to ensure that a coastline has two end points and does not coincide with each other; secondly, based on coastline vector data of two time phases, according to a specific spatial analysis method, automatically obtaining a gap region, automatically giving two attributes of sedimentation and erosion according to a certain rule, carrying out consistency check and precision evaluation, and finally obtaining the distribution range and area of sedimentation and erosion. The method and device can automatically extract erosion and sedimentation areas, reduce misjudgment, erroneous judgment and missed judgment caused by artificial interpretation, improve the detection speed, and quickly and accurately obtain the dynamic change of coastal zone intertidal zone information.

The invention discloses a vehicle violation judgment method, device and equipment and a computer readable storage medium, and the method comprises the steps: carrying out the real-time tracking of a target vehicle when a first target vehicle is detected to be located in a first region of an intersection real-time image and a second target vehicle is detected to be located in a second region; whenthe first target vehicle turns left to enter a turning penalty area, obtaining a first evidence map; when the first target vehicle blocks a second target vehicle going straight to the turning penaltyarea in the left turning penalty area, obtaining a second evidence map; when the first target vehicle crosses a stop line located at the left intersection of the turning penalty area, obtaining a third evidence map; and according to the first evidence map, the second evidence map and the third evidence map, determining that the first target vehicle has a violation behavior of turning left withoutgoing straight. According to the technical scheme disclosed by the invention, the automatic judgment that the left turn does not make straight movement is realized, so that the human input and missedjudgment conditions are reduced.

The invention provides an in-orbit autonomous health management system and management method for a long-life power supply system for space. The system comprises a health management unit, a data transmission computer and a power supply system ground fault diagnosis system, the health management unit comprises a control subsystem self-detection unit, a sun-facing orientation subsystem self-detectionunit and a power manager. The autonomous health management system can ensure that faults can be found and diagnosed in time outside a measurement and control area, astronauts can carry out disposal work in real time, fault information in the measurement and control area falls to the ground, and deep troubleshooting and detailed disposal can be carried out; a traditional human-oriented ground monitoring mode is changed, faults can be autonomously detected and diagnosed through the design of the autonomous health management system, the phenomena of misjudgment and missed judgment are reduced, the burden of ground flight control personnel is relieved, and long-term stable operation of a power supply system is maintained.

Disclosed is an electron beam focusing device for an X-ray tube. A filament, an electron beam focusing regulating unit and an anode target are arranged in an X-ray seal pipe barrel, a vacuum pump is connected with the X-ray seal pipe barrel through a pipeline, a high voltage power supply unit is connected with the X-ray seal pipe barrel through a high voltage cable, and an electronic control unit is connected with the electron beam focusing regulating unit. The focusing device integrates the modern control technique, the modern material technique and the modern nondestructive testing technique to focus electron beams so as to enable the focal spot effects to achieve approximate point light source effects. When the device is used in X-ray nondestructive testing, the X-ray image resolution is higher during imaging, therefore negative judgment or misjudgment for defects of articles to be tested can be reduced obviously according to the image effects, and the inspection effects for the articles to be tested are more truthful.

The invention discloses a programmed withstand-voltage test bench which comprises a rack, and the inside of the rack is provided with a bench body structure and a scanner, the bench body structure is used for placing an electric energy meter, and the scanner is used for scanning a bar code of the electric energy meter; the bench body structure is provided with thimble devices which are contacted with a wiring end of the electric energy meter, and each thimble device comprises a fixed block, a strong-current thimble and a weak-current thimble, the fixed block is fixed on the bench body structure, the strong-current thimble and the weak-current thimble are contacted with the wiring end of the electric energy meter, and the thimbles are arranged on the fixed block; the thimbles are connected with a high-voltage generator for generating high voltages, the high-voltage generator is controlled by a programmed circuit board, and the bar code acquired by the scanner is input into the programmed circuit board; each thimble device is corresponding to an amperemeter for measuring the leakage current of electric energy meter at high voltage, and the amperemeter is connected with the programmed circuit board; and the programmed circuit board is connected with an upper computer by a communication module. The programmed withstand-voltage test bench disclosed by the invention has the advantages of high detection efficiency, and capability of reducing the occurrence of misjudgment and judgment missing, guaranteeing the product quality, and avoiding the occurrence of high-voltage electric shock accidents.

The embodiment of the invention provides an image processing method and device, diagnosis equipment and a computer storage medium. The method comprises the following steps: the diagnosis equipment acquires a to-be-diagnosed picture of the detection object, and determines a first diagnosis result according to the to-be-diagnosed picture, wherein the first diagnosis result comprises a to-be-diagnosed target existing in the to-be-diagnosed picture and a first diagnosis value corresponding to the to-be-diagnosed target; then, the diagnosis equipment acquires a local picture comprising the to-be-diagnosed target from the to-be-diagnosed pictures, and determines a second diagnosis result according to the local picture, wherein the second diagnosis result comprises a second diagnosis value corresponding to the to-be-diagnosed target; therefore, the target diagnosis result of the to-be-diagnosed picture is determined according to the first diagnosis result and the second diagnosis result, andthe target diagnosis result comprises the to-be-diagnosed target and the diagnosis value corresponding to the to-be-diagnosed target, so that the accuracy during focus detection can be effectively improved, and missed judgment or misjudgment is reduced.

The embodiment of the invention discloses a satellite internet text sensitive information detection method and device based on deep learning. The satellite internet text sensitive information detection method based on deep learning comprises the steps: carrying out sensitive word detection on a target text, and determining sensitive words in the target text, wherein the target text is transmittedthrough the satellite internet; based on the context of the sensitive word, determining an emotional tendency type of the target text by utilizing a trained FastText model, wherein the emotional tendency type comprises a first tendency type and a second tendency type; and based on the target text of the second tendency type, generating the target text abstract by using a trained neural network model. According to the embodiment of the invention, sensitive word detection and emotional tendency classification are combined, misjudgment on the target text is reduced, and the target text with relatively long content is changed into the target text abstract with relatively short content through the neural network model, so that the accuracy and efficiency of sensitive information detection are effectively improved.