Patents
Literature
Eureka-AI is an intelligent assistant for R&D personnel, combined with Patent DNA, to facilitate innovative research.
Eureka AI

114results about How to "Improve the detection effect" patented technology

Real-time human body detection method based on AdaBoost frame and colour of head

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

Human face anti-counterfeiting detection method of double-channel convolutional neural network based on attention model

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

Railway foreign matter invasion detection method based on railway monitoring

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

Optimization method of parallel pooling layer for optimizing surface abrasion detection model of pantograph carbon contact strip

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

Tire X-ray defect detection method based on twin distance comparison

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

Multi-target detection method and device and mobile terminal

ActiveCN109902697AFlexible and efficient detection methodImprove the detection effectCharacter and pattern recognitionEngineeringConvolution
The invention provides a multi-target detection method. The multi-target detection method comprises the steps that a preprocessing module carries out convolution and pooling operation on a to-be-detected image; The first dense connection module performs convolution operation on the output of the preprocessing module and then performs series operation with the output of the preprocessing module; The first transition module performs convolution and pooling operation on the output of the first dense connection module; The second dense connection module performs convolution operation on the outputof the first transition module and then performs series operation with the output of the first transition module; The second transition module carries out convolution operation on the output of the second dense connection module; The third transition module performs pooling and convolution operation on the output of the first transition module; The extraction feature layer carries out convolutionoperation and residual module processing on the output of the first transition module and the series connection result of the output of the second transition module and the output of the third transition module; The prediction layer processes the output of the extraction feature layer and decodes the predicted target position; And the non-extreme value suppression module carries out post-processing on the output of the prediction layer.
Owner:SPREADTRUM COMM (TIANJIN) INC

High-resolution polarization SAR ship target and interference target identification method

The invention discloses a high-resolution polarization SAR ship target and interference target identification method, and belongs to the field of SAR image processing. The method is used for removing a false alarm caused by an interference target suspected as a ship in the SAR image ship target detection process. The relation among the polarization information of a ship target, the scattering intensity peak value and the structure is utilized comprehensively, Krogager polarization decomposition is carried out on a target section extracted through a ship target detecting algorithm, then a target complex structure area, namely, a target peak value area, is obtained through the changing curve of the spirochete scattering proportion and the peak value area through calculation, the spirochete scattering proportion of the peak value area is extracted, identification statistical magnitude is formed, judgment is carried out through an experience threshold, and an identification result is obtained. The method has good identification capacity on the ship target in the high-resolution polarization SAR image and the interference target suspected as the ship, the false alarm caused by the interference target in the SAR image ship target detection process can be reduced, and the utilization and popularization value is achieved.
Owner:NAVAL AERONAUTICAL & ASTRONAUTICAL UNIV PLA

Fire-fighting unmanned aerial vehicle image fusion method based on visible light and infrared thermal imaging

The invention discloses a fire-fighting unmanned aerial vehicle image fusion method based on visible light and infrared thermal imaging. The method comprises the steps: using a visible light camera and an infrared thermal imaging camera of a fire-fighting unmanned aerial vehicle to capture a visible light image and an infrared thermal image respectively; converting the two pictures into an image with the size of 1280 * 720 through a size function in the OpenCV; extracting ROI vertex coordinates in the infrared thermogram and displaying the ROI vertex coordinates in the visible light image by using a red rectangular frame; fusing an image in the edge obtained in the infrared thermogram and an image of a coordinate corresponding to the visible light by using a Laplace pyramid algorithm to obtain fire point information; and after the fire point is detected, calculating the three-dimensional position of the target according to the visual information, and accurately estimating the three-dimensional position information of the fire point under the coordinate system of the fire-fighting unmanned aerial vehicle by calculating the relative height and yaw angle between the unmanned aerial vehicle and the target. The method has the advantages of being simple in structure, convenient to operate and easy to implement.
Owner:GUANGDONG UNIV OF TECH

Downward-looking three-dimension and downward-looking multi-beam integrated underwater panoramic three-dimensional imaging sonar

The invention discloses downward-looking three-dimension and downward-looking multi-beam integrated underwater panoramic three-dimensional imaging sonar. The sonar comprises downward-looking multi-beam sonar, downward-looking three-dimensional sonar, an underwater control module and a towed body frame, wherein the towed body frame is of an integral flat structure, a downward-looking three-dimensional sonar receiving array is arranged at the bottom of the towed body frame, and a downward-looking multi-beam sonar transmitter and a downward-looking multi-beam sonar receiver are arranged in frontof and at the back of the upper part of the downward-looking three-dimensional sonar receiving array; a downward-looking multi-beam sonar transmitting array and a downward-looking multi-beam sonar receiving array are arranged in the middle of the front of the towed body frame; a downward-looking three-dimensional sonar transmitting array, the underwater control module, a downward-looking three-dimensional sonar transmitter and a downward-looking three-dimensional sonar receiver are arranged at the back of the towed body frame; transmitting and receiving electronic cabins are further arranged in the towed body frame. The downward-looking multi-beam sonar and the down-looking three-dimensional sonar are integrated, buried objects are detected in a panoramic three-dimensional manner, and three-dimensional imaging of the seabed and terrain imaging of the seabed are performed in real time, and equipment exploration capability and detection efficiency are improved.
Owner:中科探海(苏州)海洋科技有限责任公司

Double-parameter CFAR (Constant False Alarm Rate) ship detection method based on Fourier transformation

The invention discloses a double-parameter CFAR (Constant False Alarm Rate) ship detection method based on Fourier transformation. The method comprises the steps of acquiring SAR (Synthetic Aperture Radar) radar echo data, and imaging the SAR radar echo data to obtain an SAR radar imagingdata matrix I, and further obtain a binary image matrix Ibw subjected to threshold processing, wherein the Ibw is a binary image containing a ship target and a false target; setting an all 1 matrix as a background window; sequentially calculating a clutter number statistical matrix N in the background window and a clutter number statistical matrix in the background window after reciprocal taking operation, and further calculating a clutter mean value statistical matrix M in the background window; then calculating a clutter variance statistical matrix V in the background window, and calculating a ship target determination matrix F; finally detecting to obtain a plurality of ship targets according to I and F, wherein each detected ship target is an interesting ship target, so that the separation of the interesting ship targets and false targets is accomplished, and the detection of the interesting ship targets is further accomplished.
Owner:XIDIAN UNIV

Fluorescence immunochromatography detection method and device based on sparse self-coding neural network

The invention relates to a fluorescence immunochromatography detection method and device based on a sparse self-coding neural network. The device comprises a photoelectric signal detection unit, a mechanical scanning unit, an STM32 microprocessor control unit, and a computer in which a fluorescence detection sparse self-coding neural network module and a fluorescence immunochromatography detection database are established, wherein the mechanical scanning unit controls the movement of a fluorescence test strip; the photoelectric signal detection unit detects the fluorescence intensity change of the moving fluorescence test strip and converts the fluorescence intensity change into electrical signal data; then the electrical signal data is transmitted to a computer through the STM32 microprocessor control unit, so as to be subjected to filtering processing, and background interference of a base line is eliminated; data analysis is performed through the fluorescence detection sparse self-coding neural network module and the fluorescence immunochromatography detection database is performed, so that a fluorescence immunochromatography detection result is obtained. According to the method, characteristic value extraction of detection data is avoided, data can be directly detected, character representation of data is learnt in a layered mode, and the fluorescence immunochromatography detection effect is improved.
Owner:FUZHOU UNIVERSITY

Inspection device used for production of high voltage vacuum circuit breaker

The invention discloses an inspection device used for production of a high voltage vacuum circuit breaker, comprising a shell, wherein a plurality of insulated sleeves are sequentially arranged at thetop of the shell from left to right in sequence, a wiring terminal is arranged at the top end of each insulated sleeve, a breaking-closing operation handle is movably connected to the right side of the front part of the shell, an insulated handle is fixedly connected to the top end of the breaking-closing operation handle, an insulated handle sleeve is sleeved on the left side of the insulated handle, and an insulated pressure lever is fixedly connected to the center of the bottom of the insulated handle sleeve. According to the invention, by the match of the insulated handle sleeve, the insulated pressure lever, a fixed rack, a sliding sleeve, a fixed transverse plate, a bearing seat, a movable contact, a fixed seat, a static contact, a pressure sensor, a junction box, a wire and a control panel, pressure values in high voltage vacuum circuit breaker breaking and closing can be accurately measured, and therefore the detection effect of the detection device is improved, the problem ofpoor detection effect of conventional detection devices is solved, and the percent of pass of the high voltage vacuum circuit breaker can be increased.
Owner:许昌市长江高压计量设备有限公司

Self-adaptive detection method for synchronism of lamplight/motor and sound

The invention relates to a self-adaptive detection method for synchronism of a lamplight/motor and sound. The self-adaptive detection method comprises the following steps: detecting a phonetic acoustic signal by an ADC (analog to digital converter) controlled by AGC (automatic gain control); outputting a narrow dynamic range phonetic signal with AGC information; restoring the narrow dynamic range phonetic signal with the AGC information into a wide dynamic range phonetic signal by an energy normalization algorithm; carrying out phonetic real-time energy calculation and threshold calculation by the restored phonetic signal; and through comparison of the real-time energy and the threshold, calculating a motor starting and stopping state or a lamplight turning on and off state corresponding to the current phonetic frame. Through the introduction of the AGC information, high dynamic range obtained on the ADC with low sampling depth is detected by the motor opening and closing state or the lamplight turning on and off state information; and with the adoption of the ADC with 12 bit low sampling depth, the motor opening and closing state or lamplight turn on and off state when users speak can be detected at the distance of 5-300cm, so that the overall cost of the system is greatly reduced.
Owner:HEFEI IFLYTEK TOYCLOUD TECH

Formaldehyde-removing building material inspection device

The invention provides a formaldehyde-removing building material inspection device, and relates to the field of building material inspection. The formaldehyde-removing building material inspection device comprises an inspection box, a crushing mortar and a crushed pestle; the interior of the inspection box is hollow, and a buckle fixing frame is arranged on the outer wall of the inspection box; asealing plate is slidably connected into the top of the inspection box, and a sealing cover plate is connected into the top of the sealing plate; a containing vessel is located in the inspection box;the crushing mortar is connected into a buckle fixing frame on the outer side of the inspection box in a buckled mode; tension springs are arranged at two positions, the left end and the right end ofthe crushing pestle are connected with the left end and the right end of the crushing mortar through the tension springs respectively; according to the building material inspection device, blocky building materials in the crushing mortar are repeatedly smashed through the crushing pestle, the blocky building materials in the crushing mortar are forced to be further crushed and refined, the crushedbuilding materials are detected, the material materials contained in the building materials can be fully exposed in the air, and the building material detection effect is improved.
Owner:许霞

Method for clustering image areas through LLC based on adaptive codebook

The invention relates to a method for clustering image areas through LLC based on an adaptive codebook. Color, texture, Gabor, center of mass and other features are extracted from the segmented super-pixels and then the neighbor areas are clustered through the fused features, the identifiable degree between the multiple neighbor areas is enhanced by the form of a similarity structure diagram, the extremely valuable clues are provided for reducing and even eliminating the noise near the boundary of the foreground in the generated saliency map and the boundary of the foreground is enabled to be clearer; multiple features are used as the basis for calculating the saliency map, and when the color feature cannot be used for effectively extracting the saliency target under the complex scene, multi-feature information is used as the beneficial supplement to effectively enhance the detection result; and as the extended LLC coding scheme, the method that multiple feature descriptors in the original LLC are coded and then fused in the objective function is extended into the method that the multiple feature descriptors are fused firstly and then coded in one step so that the coding process can be simplified and the integrity of the multiple features can be emphasized.
Owner:HENAN UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products