131results about How to "Accurate automatic identification" patented technology

The invention discloses an image-processing-based unmanned plane accurate position landing method, which comprises the following steps: (1) a GPS (global position system) satellite navigation system enables an unmanned plane to be located above a ground parking apron; (2) an air pressure height measurement gauge and a distance measurement module of an ultrasound radar are combined to control the ground clearance for the unmanned plane to land; (3) a vision module identifies a coarse positioning identification domain in real time, and combines Hough Transform and RGB mean value method and gate position identification to process a coordinate of a targeted landing point; (4) when landing of the unmanned plane meets the threshold condition of the coarse positioning identification domain, the algorithm in the step (3) is utilized to perform accurate positioning treatment on the accurate positioning identification domain; and (5) the unmanned plane is controlled for accurate landing by taking the treated deviation value as the input quantity and adopting the double PID algorithm. According to the invention, the defect that insufficient GPS accuracy of the unmanned plane causes a fault landing is overcome, the intelligence for the unmanned plane control is improved, and the cost for using an accurate sensor is greatly reduced.

Disclosed is an authentication method of a dynamic two-dimension code instruction. A dynamic two-dimension code token comprises a collection module, a display module, an authentication module, an instruction module, a management module, a two-dimension code module and a control module. The control module, the management module, the authentication module, the instruction module and the two-dimension code module are integrated in the same microprocessor. The collection module is connected with the control module and the two-dimension code module. The display module is connected with the control module and the two-dimension code module. An authentication system of the dynamic two-dimension code instruction is the integration of service programs for carrying out the functions of authentication of the dynamic two-dimension code instruction, token synchronization, and management and configuration of relevant states of the token. The dynamic two-dimension code instruction has the characteristics of preventing exposure, being high in intensity, unpredictable, resistant to replay and disavowal, free of memory, accurate and efficient, and having physical isolation and bidirectional authentication, and safety of identity authentication is improved remarkably.

The invention provides a method for enhancing a pavement image, and relates to a pavement crack detecting method based on an active contour model. The pavement crack detecting method based on the active contour model comprises the following steps of: enhancing the contrast of a pavement crack image; carrying out accurate detection of a pavement crack area on the image with an enhanced contrast; and classifying pavement cracks based on an accurate detection result. According to the technical scheme provided by the invention, various crack images obtained in the high-speed operation process in a natural scene can be automatically identified. With the adoption of the method provided by the invention, the high-accuracy positioning of a crack suspicious area is realized, and the robustness identification is carried out on an area with a positioning error by using various sample classification methods.

The invention discloses an extraction method of a visual saliency area based on a monocular depth map. The extraction method of the visual saliency area comprises the following sequential steps: dividing an original image to obtain super pixels; building a characteristic vector of each super pixel to estimate an absolute depth characteristic of absolute depth in a scene in the image; building a probability model by using a Gaussian-Markov random field model, calculating a distance relation between the super pixel characteristic vector and an adjacent super pixel characteristic vector by virtue of the probability model to obtain a relative depth characteristic on the basis of the absolute depth characteristic and obtain depth values and depth maps of the super pixels simultaneously; calculating saliency values of the super pixels; calculating a gain coefficient according to the depth values and correcting the saliency values by using the gain coefficient. The extraction method can be used for quickly and automatically identifying the saliency objects in the image without any prior knowledge, is high in universality and can also be used for accurately detecting the saliency area.

The embodiment of the invention provides a method and device for automatically identifying abnormal seismic traces. The method includes the steps that true values of seismic trace sampling points are obtained from collected single-explosion seismic records, and a maximum value threshold parameter value is set according to the true values of the seismic trace sampling points; absolute values of the true values of the seismic trace sampling points are obtained and compared with the maximum value threshold parameter value; when the absolute values of the true values of the seismic trace sampling points are larger than the maximum value threshold parameter value, the seismic traces where the sampling points are located are the maximum-value-type abnormal seismic traces. By means of the technical scheme, the accuracy for identifying the abnormal seismic traces can be greatly improved.

The objective of the invention lies in providing a road condition information processing method and equipment. Specifically, the method comprises the steps: obtaining a road surface image, and recognizing a suspected road condition event through the road surface image; obtaining vehicle running information when the suspected road condition event happens; carrying out the further confirmation of the suspected road condition event according to the vehicle running information, and obtaining a reliable road condition event. Because the direct automatic recognition of the road condition event through the road surface image is liable to be wrong, the method further employs vehicle running information. For example, the above vehicle running information is very likely to happen when there is a foreign matter if the severe braking and anti-collision early warning information is obtained at the same time, thereby indicating that the possibility that there is a foreign matter at a road segment is high, and confirming that the suspected road condition event is a reliable road condition event based on the above. Therefore, the method can automatically recognize the road condition event quickly and accurately in a driving process, cannot distract the attention of a driver, and improves the driving safety.

The invention provides a motion control method suitable for exoskeleton robots. By establishing a seven-link kinematics model, collecting exoskeleton-wearer system motion information based on a multi-sensor system, calculating the wearer's ground pressure center position, and designing an exoskeleton walking finite state machine, the wearer's walking motion intention and the state transitions in the walking finite state machine can be quickly, accurately and automatically recognized, effective and smooth motion assistance for paralyzed patients who have no perception and strength of the lowerlimbs is realized, the walking intention of the exoskeleton robot is automatically recognized and controlled, and the convenience and intelligence of using the exoskeleton robots are improved. At thesame time, by calculating the distance between the ankle joint of the front leg and the location of the crutches after the exoskeleton-wearer system moves forward with both crutches during the leg support period, the step length of the wearer under different walking conditions is determined, the exoskeleton joint motion reference trajectory is updated in real time, and real-time adjustment and control of the wearer's walking step length is realized.

The invention discloses a transformer winding state recognition method based on a fuzzy adaptive resonance neural network. The transformer winding state recognition method comprises the steps of (1) arranging monitoring points of a vibration sensor, setting the sampling frequency and the sampling time of a data acquisition instrument, and acquiring transformer vibration signals; (2) performing wavelet packet decomposition and reconstruction on the transformer vibration signals; (3) extracting vibration signal sub-band energy values; (4) analyzing the vibration signal sub-band energy values of each vibration monitoring point, and selecting feature band energy of valid monitoring points to construct a feature vector so as to act as input of the fuzzy adaptive resonance neural network; (5) building the fuzzy adaptive resonance neural network, adjusting network parameters until set precision is reached; and (6) performing recognition on transformer winding state through the fuzzy adaptive resonance neural network. The transformer winding state recognition method can judge a transformer winding pressing state accurately and quickly, and can be applied to online monitoring and recognition for the transformer winding pressing state.

The invention relates to the field of exception monitoring, and provides an abnormal event monitoring method and device, computer equipment and a storage medium. The method comprises the following steps: receiving event data; performing association event analysis and classification analysis on the event data through a CEP engine model to obtain a plurality of basic index values; performing event classification on the event data through an event management model to obtain event types, and performing event analysis and event analysis on the event data through the event management model to obtaina plurality of potential index values; obtaining an event prediction model and an abnormal threshold matched with the event type; performing combination and abnormal feature extraction on all the basic index values and all the potential index values through an event estimation model to obtain an estimation result; and determining a monitoring result of the event data according to the estimation result and the abnormal threshold. According to the invention, whether the event data is the abnormal event or not can be accurately, comprehensively and automatically identified, and the security of the event data is ensured.

The invention provides a method for identifying an infrared movement small target based on significance and structure, and belongs to the application technical fields such as computer vision, pattern recognition and image processing. In order to identify a small movement target from an infrared scene sequence image under the condition of no priori knowledge, the target identification method based on the significance and structure is provided. The method comprises the following steps: firstly, performing significance detecting to an image of each frame as per the significance characteristics; secondly, excluding non-targets through the structure of the movement target; performing the automatic identification of the target through a continuous track of the movement target. According to the invention, the method can be used for identifying the infrared small target quickly and precisely in the premise that no priori knowledge is provided.

The invention discloses a license plate classification method and system based on deep learning, and is applied to an electronic apparatus. The method comprises the following steps: obtaining at least one picture sent by a terminal device; preprocessing the obtained at least one picture so as to enable the preprocessed at least one picture to match input parameters of a pre-trained identification model; and inputting the preprocessed at least one picture into the pre-trained identification model, identifying vehicle purpose information corresponding to the at least one picture by use of the pre-trained identification model, and sending the vehicle purpose information corresponding to the at least one picture to the terminal device. The purpose of a vehicle in the picture can be accurately and automatically identified, accurate classification can also be realized, and the accuracy and the work efficiency are improved.

The invention discloses an ultrasonic automatic detection system of copper and stainless steel dissimilar metal vacuum brazing water connector, a hardware of which consists of a water connector automatic detection system, a single crystal immersion focusing probe and a reference block; the detection device includes an industrial control machine, a motor driver and motors which are controlled by the industrial control machine and an ultrasonic flaw detection instrument which is respectively connected with the single crystal immersion focusing probe and the industrial control machine; the invention is characterized in that: one motor of the detection device drives a transmission shaft which is moved vertically, a probe angle adjuster is arranged on a shaft, the single crystal immersion focusing probe is arranged at the lower end of a probe angle adjuster, the other motor drives a horizontal rotating disk which is provided with the water connector to be detected, the position of the probe in a water connector cavity body can be regulated by the regulation of the probe angle adjuster, and the ultrasonic waves are transmitted for the detection of the brazing quality of the water connector. The invention has the advantages that the probe structure is compact, the detection system realizes the automatic control, the detection sensitivity is high, the operation is simple, and the invention can automatically identify the defect, calculate the area of defect and provide the spreading picture of the detection results of the water connector rapidly, accurately and intuitively.

The invention discloses an adhered piglet automatic counting method based on ellipse fitting, and belongs to the field of image processing. The problem of adhered piglet automatic counting is solved. The method comprises the implementation steps that (1) an adhered piglet gray level image is opened and preprocessed; (2) the edges of the communicated areas of the piglet image are extracted, and ellipse fitting is executed; (3) the communicated areas which are not processed are extracted in an area descending order, if ellipses corresponding to the communicated areas which are not processed meet the parameter requirement of a single piglet image fitting ellipse, the communicated areas are marked to be processed, otherwise, the contour line of the communicated areas is segmented based on concave points, ellipse fitting is carried out on each segment, a plurality of ellipses belonging to the same piglet are combined according to a proposed combination rule, and the communicated areas are marked to be processed after combination is finished; (4) if the communicated areas which are not processed still exist, the step (3) is executed, and otherwise, counting is finished, and the number of the ellipses is the number of the piglets. According to the method, the number of the piglets in the adhered piglet gray level image can be recognized automatically and accurately, and the ellipses can reflect the rest behavior characteristics of the piglets.

A full-automatic tungsten ore concentrator comprises a feeding system which is used for vibrating feeding a plurality of ores so as to enable the plurality of ores to drop and sequentially pass through a shooting area and a spraying area; setting a camera device arranged in the shooting area to photograph the ores; a spraying valve arranged in the spraying area and used for jetting compressed airto the falling ores; an image processing system which is used for receiving an ore photograph and processing the ore photograph so as to find the tungsten ores and the spatial coordinates of the tungsten ores; a control system which is used for calculating the falling time of the tungsten ores falling at the position of the spraying valve based on the space coordinates of the tungsten ores, and controlling the spraying valve to inject compressed air into the tungsten ores for sorting on the basis of the falling time.

The invention discloses an excavator working state identification method based on a hybrid LBF (Local Binary Features) shape regression model. The method comprises the following steps: 1) training a hybrid LBF shape regression model of an excavator, and using the model to predict the shape (i.e., a set of relative coordinates of feature points) of the excavator input to a video frame; 2) calculating working state feature descriptors of the excavator according to the coordinates of the feature points and the detected length-width ratio of the excavator; and 3) utilizing a SVM classifier to determine the current state (that is to say, a working state or a non-working state) of the excavator. The excavator working state identification method accurately and automatically identifies the working state of the excavator on a land, and provides an intelligent means for on-site construction monitoring of a construction site.

The invention relates to a garlic planter. The garlic planter is characterized by comprising a garlic seed box, a flywheel, a primary differentiation plate, a conveyor belt, conveying troughs, a control module, a detection module, a support baffle plate, a garlic clove support device, collection funnels and a planting device, wherein the flywheel is mounted at the bottom of the garlic seed box; the primary differentiation plate is mounted below the flywheel; the conveyor belt is mounted below the primary differentiation plate; the conveying troughs are formed below the conveyor belt, and the detection module for performing information collection on garlic cloves on the conveying troughs is arranged above the conveying troughs; the collection funnels are correspondingly arranged below the conveying troughs; the garlic clove support device is mounted between the conveying troughs and the corresponding collection funnels; the collection funnels are connected with the garlic clove planting device; the conveying troughs is are formed in a conveying mechanism; and the control module is connected with the detection module and a driving mechanism respectively. The garlic planter has the advantages of being capable of accurately and automatically identifying the orientation of the garlic cloves and garlic buds and can realize the direction adjustment and vertical planting of the garlic cloves, and is high in degree of automation, stable and efficient.

The invention discloses a road traffic sign recognition method, and the method comprises the steps: constructing a deep convolution neural network, selecting a traffic standard database and a trafficsign image collected on site, carrying out the training and testing of the constructed deep convolution neural network, and obtaining the trained deep convolution neural network; collecting a real-time traffic sign image; adding a traffic sign position distribution priori knowledge to an Itti model, so as to adjust the collected traffic sign image; carrying out the brightness equalization and color enhancement, and then extracting an AOI of the traffic signal through an MSER algorithm; inputting the AOI into the trained deep convolution neural network for convolution and pooling processing, and obtaining a one-dimensional feature vector; completing the recognition of the one-dimensional feature vector through a full-connection BP neural network, outputting a recognition result, and carrying out the real-time display and broadcasting. According to the invention, the method can accurately recognize various types of collected traffic signs, and achieves the timely prompt of a driver through texts or voice so that the driver takes corresponding action.

The invention discloses a hazardous article three-dimensional image detection method and device, computer equipment and a storage medium, and the hazardous article three-dimensional image detection method comprises the steps: receiving a hazardous article detection instruction, and obtaining a to-be-detected three-dimensional image, wherein the to-be-detected three-dimensional image is a reconstructed three-dimensional image after the to-be-detected object is scanned by security inspection equipment; performing 3D image segmentation on the to-be-detected three-dimensional image to obtain a plurality of square images; scanning and standardizing the target areas of all the block images to obtain a target area image; extracting hazardous article features in the target area image through a hazardous article detection model, and generating a feature vector graph; and obtaining an identification result output by the hazardous article detection model in a classified manner according to a K-means clustering algorithm. According to the hazardous article three-dimensional image detection method, segmentation, scanning and standardization processing are carried out on the three-dimensional image, and then classification and identification are carried out according to the K-means clustering algorithm, so that hazardous articles can be identified accurately, rapidly and automatically, and the satisfaction degree of customers is improved.

The present invention discloses a method for monitoring efficient and safe operation of a coal mining face ribbon conveyer. The method comprises: step 1, arranging hardware equipment used for monitoring; step 2, performing data collection and transmission; and step 3, performing data processing and efficient safe and operation control of the ribbon conveyer. The method in the present invention method is simple in steps. Feature point data of a coal flow outer contour is detected by using a laser radar, a real-time belt speed of the conveyor belt is detected by using a rotary encoder, and a coal flow can be efficiently and accurately detected through data analysis processing and calculation on data by a computer. A visual sensor is used for efficiently and accurately detecting a coal contour on the conveyor belt, so that accurate and automatic recognition of large-sized coal on a mining face is implemented. Coal flow detection efficiency is high, and precision is high. A large-sized coal recognition method is simple and reliable, has a small error, is of great significance to efficient and safe operation of an underground coal mine ribbon conveyer, and is easy to popularize and use.

The invention relates to the technical field of big data analysis, and provides a low-voltage transformer area topology identification method and a low-voltage transformer area topology identificationdevice. A low-voltage transformer area comprises a plurality of line branches. The low-voltage transformer area topology identification method comprises the steps of: collecting power utilization information data of each line branch at a preset time interval in a preset time period, and forming an original data matrix; preprocessing the original data matrix to obtain a standardized data matrix; mapping the standardized data matrix into a low-dimensional space of a preset dimensionality to obtain a low-dimensional data matrix; and drawing a topological structure of the low-voltage transformerarea according to the low-dimensional data matrix. According to the low-voltage transformer area topology identification method and the low-voltage transformer area topology identification device provided by the invention, the topological structure of the low-voltage transformer area can be accurately, efficiently and automatically identified, so that the management level of the low-voltage transformer area is greatly improved.

The present invention provides a diagnosis system suitable for different vehicles, wherein the vehicle comprises vehicle model information and a plurality of control systems corresponding to the vehicle model information. The diagnosis system comprises a diagnosis device; the diagnosis device is used for sending a call instruction to a vehicle to be tested, calling the vehicle model information of the vehicle according to the call instruction and determining the plurality of control systems corresponding to the vehicle model information according to the called vehicle model information; and the control systems are connected with the gateway of the vehicle and are used for transmitting the state information and communication information of the vehicle to the diagnosis device so that the diagnosis device realizes the fault detection of the vehicle. With the diagnosis system of the invention adopted, the problem of need for a large amount of money to realize diagnosis software upgrade after the configurations of new vehicle models on the market or sold vehicles models are modified can be solved, the maintenance period of after-sales diagnosis software can be shortened, and the stability of the after-sales diagnosis software can be ensured.

The invention provides an ore rapid identification method which comprises the steps of (S1) acquiring an ore training image and training the ore training image based on depth learning to obtain a plurality of modeling coefficients, (S2) building a processing model based on the plurality of modeling coefficients, (S3) testing an ore image to be tested based on the processing model to obtain a maskimage, and (S4) obtaining an ore image based on the ore image to be tested and the mask image to identify ore. The ore rapid recognition method and a computer readable storage medium are implemented,the processing model is obtained based on deep learning training, and the ore can be automatically and accurately identified with high accuracy.

The invention discloses a bituminous pavement loose recognition and calculation method based on a three-dimensional laser technology. The method comprises the steps: 1) acquiring pavement three-dimensional laser point cloud data of a detected road section; 2) carrying out smoothing treatment on the pavement three-dimensional laser point cloud data; 3) based on the laser point cloud data after the smoothing treatment, establishing the cross-section morphology of a benchmark pavement and the cross-section morphology of a loose pavement; 4) uniformly dividing the cross section of each pavement into M division sections with equal distance; 5) calculating characteristic parameters of the M division sections of each cross section; 6) determining whether the cross section of each pavement is a loose cross section one by one, 7) calculating the width and depth of each loose cross section; 8) identifying a loose zone; and 9) calculating the length, width, depth and area of the loose zone. The method realizes the automatic identification on loose, not only is rapid and accurate, but also can avoid the dangerousness of worker on-site exploration and prospection.

The invention relates to a fully-automatic vision BGA repairing work station. The work station comprises a rack, a fuselage outer cover and a rack platform. The two sides of the rack platform are provided with Y-axis moving mechanisms, X-axis moving mechanisms are arranged on one end of each of the two-side Y-axis moving mechanisms, a Z-axis moving mechanism is arranged above the X-axis moving mechanisms, a lower-portion double-temperature area structure is arranged below the two-side Y-axis moving mechanisms, and the lower-portion double-temperature area structure is directly installed on therack platform so as to form the fully-automatic vision BGA repairing work station. A machine adopts a computer control mode to drive the X-axis moving mechanisms, the Y-axis moving mechanisms and theZ-axis moving mechanism to complete X-axis direction, Y-axis direction, the Z-axis direction and R axis (R angle) intelligent motion system operation. Accurate alignment is realized, an absorption material and a pasting and installation height are automatically identified, and automatic alignment, automatic pasting and installation, automatic welding and automatic desoldering operation are performed so that an effect of increasing the quality of an attached SMT patch component is achieved.

The invention relates to a method for identifying the brand and model of a projector, comprising the steps of: S1, acquiring an image of an area where the brand and the model of the projector are located; S2: preprocessing the collected image to extract candidate target area; S3, screening the candidate target area by adopting a target area screen strategy to obtain a target area; S4: according tothe obtained target area, recognizing the character in the target area by using the OCR technology, and obtaining the character string of the target area; S5, matching the recognized target area string with the string data in the brand and model database to obtain a matching result of the brand and model. The invention can accurately and automatically identify the brand and model text informationon the surface of the projector fuselage shell.

The invention relates to a method for automatically identifying a waterline scale. The method includes obtaining standard sample images based on shot ship staff gauge images, identifying the water surface boundary from the standard sample images, obtaining staff gauge number images above the water surface boundary and containing staff gauge numbers, obtaining a target images through the images of each staff gauge number within four edges according to the staff gauge number images, zooming in the target images to the size of a template sample image, and calculating the matched evaluating value of each target image with the template sample image, wherein the template sample image corresponding to the maximal image matching assessment value is identified as the waterline scale. The waterline scale can be automatically and accurately identified when the staff gauge scale is not clear.

A tungsten ore recognition and sorting method comprises the following steps of s1, allowing a plurality of ores to fall off and sequentially pass through a shooting area and a spraying area; s2, a camera device arranged in the shooting area photographing the ores and sending the obtained ore pictures to an image processing device; s3, the image processing device processing the ore pictures to findout the tungsten ores and the space coordinates of the tungsten ores; s4, a control system calculating the falling time of the tungsten ores falling at the position of a spraying valve in the spraying area based on the space coordinates of the tungsten ores, and controlling the spraying valve to spray compressed air into the tungsten ores for sorting on the basis of the falling time. According tothe method for identifying and sorting tungsten ores, ore automatic recognition and sorting can be carried out. The method, compared with manual selection, is high in speed, high in production efficiency, low in cost and high in sorting rate.

A trademark image identification method provided by the present invention comprises the following steps of preparing a training sample of a to-be-identified trademark, extracting the local characteristics of the training sample, and carrying out the clustering and quantification; utilizing the mutual information to select the characteristics; extracting the local characteristics of a tested image same as the local characteristics of the training sample, and utilizing a clustering center obtained previously to quantize the local characteristics; utilizing the characteristics retained by the characteristic selection to filter the key points of a tested picture, and carrying out the key point matching on the tested picture and a positive sample; utilizing a spatial topological relation to restrain the matching, removing the key point pairs of the wrong matching, and gathering the number of different visual words in the point matching pairs as the similarity of the tested image and the positive sample; and according to the similarity, determining whether the tested image contains the trademark. The trademark image identification method of the present invention considers from the two aspects of irrelevant characteristic point filtering and the wrong matching elimination, the irrelevant characteristic point filtering and the wrong matching elimination are the complementary relation and can be promote mutually, thereby improving the trademark identification accuracy better.

The embodiment of the invention provides a method and system for identifying vehicle appearance. The method comprises the steps: photographing a plurality of current driving images of the vehicle through a plurality of cameras in a predetermined monitoring area; based on the plurality of current driving images, identifying identity information of the vehicle; according to the identity information of the vehicle, performing image splicing processing on a plurality of current driving images of the vehicle to obtain a current appearance image of the vehicle; and carrying out vehicle appearance analysis on the current appearance image of the vehicle to obtain an analysis result, and determining whether the appearance of the vehicle is damaged or not according to the analysis result. Through the method, under the condition that a plurality of vehicles run in a preset monitoring area at the same time, the current running images of the vehicles can be identified efficiently and accurately, and the current appearance images of the vehicles are spliced, so that whether the appearances of the vehicles are damaged or not can be identified efficiently, accurately and automatically, and the use experience of vehicle users is greatly improved.

The invention relates to a radio frequency identification (RFID) technology-based gun management system, which consists of a computer, a gun borrowing-returning module, a gun examination module and an RFID electronic label which is arranged on a gun. The gun borrowing-returning module comprises a desktop-typed RFID reader-writer which is connected with the computer through an Ethernet interface; and the gun examination module comprises a handheld RFID reader-writer which is communicated with the computer through a wireless network. The RFID technology-based gun management system has the advantages that: (1) real-time mandatory administration can be realized; (2) the system is convenient to use, and the labor workload can be reduced; and (3) the safety is good, and the system is difficult to copy.