1421results about How to "Reduce missed detection rate" patented technology

The invention relates to the technical field of mobile robot navigation and discloses a mobile robot cascading map building method based on remarkable scenic spot detection. The mobile robot cascading map building method comprises the steps of: 1) according to image data collected by a mobile robot sensor, carrying out on-line detection on a road sign of a natural scene corresponding to a remarkable scene so as to generate a topological node in a global map; 2) updating the pose of a mobile robot and a local grid map; and 3) building a global topological map structure by taking a remarkable scenic spot as the topological node, and optimizing the topological structure by introducing a weighted scanning and matching method and a relaxation method on the basis of the closed detection of trajectory of the robot to ensure the global consistency of the topological map. The mobile robot cascading map building method is applicable to the autonomous path planning and navigation application for various mobile robots in large indoor environment ranges such as a plurality of rooms and corridors.

The invention discloses a remote obstacle detection method based on laser radar multi-frame point cloud fusion. A local coordinate system and a world coordinate system are established, an extraction feature point of each laser point is calculated on an annular scanning line of the laser radar according to the original point cloud data under the local coordinate system, and the global pose of the current position relative to the initial position and the de-distortion point cloud in the world coordinate system are obtained through inter-frame feature point matching and map feature point matching; the de-distortion point clouds of the current frame and the previous frame are fused to obtain more compact de-distortion point cloud data, which is unified to the local coordinate system, then projection is performed on two-dimensional grids, and an obstacle is screened according to the height change features of each two-dimensional grid. According to the method in the invention, the problem that the detection rate of the remote barrier caused by sparse laser point clouds is low is solved, the remote barriers can be effectively detected, the error detection rate and the leak detection rateare low, and the system cost can be greatly reduced.

The invention discloses a method for quickly and accurately detecting and tracking a human face based on a video sequence, which relates to the technical field of mode identification. The method comprises the following steps of: 1, extracting a video frame image from a video stream; 2, preprocessing the video frame image, namely compensating light rays, extracting skin color areas, performing morphological processing and combining the areas; 3, detecting the human face, namely representing the human face by using Harr-like characteristics and detecting the human face by using a cascaded Adaboost algorithm with an assistant decision function; 4, establishing the characteristics of the human face, namely detecting the area characteristics of the detected human face and the shape characteristics of the edge profile of the human face; 5, tracking the human face, particularly tracking the human face by using a human face area characteristic model when an intersection does not occur in a human face area, and further matching when the intersection occurs according to the shape characteristics of the edge profile of the human face; and 6, extracting the sequence of a human face image. By the technical method, the human face can be detected and tracked quickly and accurately on the basis of the video sequence.

The invention discloses a pedestrian detection method based on an end-to-end convolutional neural network in order to solve the problem that the existing pedestrian detection algorithm has the disadvantages of low detection precision, complex algorithm and difficult multi-module fusion. A novel end-to-end convolutional neural network is adopted, a training sample set with marks is constructed, and end-to-end training is performed to get a convolutional neural network model capable of predicting a pedestrian candidate box and the confidence of the corresponding box. During test, a test picture is input into a trained model, and a corresponding pedestrian detection box and the confidence thereof can be obtained. Finally, non-maximum suppression and threshold screening are performed to get an optimal pedestrian area. The invention has two advantages compared with previous inventions. First, through end-to-end training and testing, the whole model is very easy to train and test. Second, pedestrian scale and proportion problems are solved by constructing a candidate box regression network, the pyramid technology adopted in previous inventions is not needed, and a lot of computing resources are saved.

The invention discloses a fatigue driving fusion detection method based on soft computing, which can detect the fatigue driving of the driver and is characterized in that: the fatigue driving is detected in fusion mode through two aspects which include two facial characteristics for directly indicating the fatigue state of the driver and two vehicle behavior characteristics for indirectly indicating the fatigue state of the driver, wherein the two facial characteristics respectively are frequent blinking and yawning, and the two vehicle behavior characteristics respectively are abnormal vehicle lane deviation and abnormal steering wheel rotation; the invention utilizes the TS fuzzy neural network to recognize fatigue driving, adopts abstraction clustering for the optimized recognition of the network structure, and determines the number of fuzzy rules of the fuzzy neural network and the initial values of the relevant network parameters; genetic algorithm is utilized to train and optimize the network parameters, and determine the optimum network parameters; the TS fuzzy neural network is utilized to detect the fatigue driving of the driver in real-time according to the optimum network parameters and the four fatigue characteristic parameters.

The invention discloses a pedestrian detection method based on a deep learning technology. The method comprises the steps that firstly, a binary classification model is trained through a step-by-step migration strategy on the basis of transfer learning to initialize final model parameters; secondly, pedestrian detection work is completed by adopting and modifying a currently popular and efficient Faster RCNN frame, and on the basis of the CNN characteristics of the frame, not only can images with any scales be processed, but also the detection speed is high. Compared with the prior art, the method has the advantages that the network does not need to be specially designed, existing available data is fully utilized, a good experiment result still can be achieved by adopting a general network structure, the advantages of a deep convolution network are fully achieved, and the advantages of being simple in design, good in robustness, high in detection accuracy and low in omission ratio are achieved.

The invention belongs to the field of deep learning and intelligent vehicle road detection, and specifically relates to a vehicle detection method based on improved YOLOv3. The method comprises the following steps: redesigning a convolutional neural network structure between a Darknet layer and three yolo layers, and designing a YOLO-TN network by referring to an idea of TridentNet weight sharing;carrying out model pruning on the YOLO-TN convolutional neural network; constructing a vehicle detection data set, and labeling vehicle position information in the data set; respectively training vehicle detection models based on the YOLO-TN and the YOLOv3, completing a vehicle detection task, and comparing detection results of the YOLO-TN and the YOLOv3. The method has high average precision while guaranteeing real-time detection, and is lower in omission ratio and more accurate in positioning when used for detecting distant vehicles and small-scale targets.

The invention discloses a method for detecting a random access leader. After a random access channel (RACH) signal is treated with the simplified time domain-related operation, noise mean estimation is carried out to each root sequence; an absolute threshold is then detected according to a signal obtained by the noise estimation and signals in a search window are detected; the signals which exceed the detection threshold are screened and then a timing position can be determined in a fast and accurate way so as to provide reliable uplink timing adjustment information to a mobile terminal, guarantee the accuracy of the uplink synchronization and effectively eliminate the timing error due to the over-sampling and reduce the false alarm in the detection.

The invention provides a security protection early-warning system and method based on radar detection and video image behavior analysis. The method comprises: S1, carrying out pairing and calibrationon a radar detecting device and a cloud platform camera; S2, monitoring whether a moving target exists in a monitoring range in real time by a radar and sending position information of all moving targets to an intrusion event detecting module; S3, determining positions of the moving targets relative to the cloud platform camera; S4, determining an adjustment angle of the cloud platform camera anda needed magnification value; S5, moving the visual field center of the cloud platform camera to the positions of the moving targets and controlling the cloud platform camera to carry out magnification; and S6, acquiring a current image of the cloud platform camera, carrying out a behavior analysis on the moving targets, determining a real intrusion target, and sending out alarm information. The security protection early-warning method has advantages of simple algorithm, fast recognition speed of the intrusion target, high recognition accuracy, and flexible equipment installation position arrangement and the like; the missed detection rate and the false alarm rate are reduced substantially; and the response speed of the system is increased.

The invention discloses a multi-target tracking method based on multi-model fusion and data association; the tracking method comprises the following steps: firstly using an interframe difference method to detect a motion target contour and center of mass coordinates; fusing a pyramid optical flow method with Kalman filtering so as to predict the center of mass coordinates of the motion target in the next moment; using Euclidean distances between the center of mass coordinate predicted value and the center of mass coordinate detection value at next moment to form a benefit matrix, and using a Hungary algorithm to obtain the optimal matching through data association; finally removing certain portion unable to satisfy requirements in a tracker, and building a tracking unit for non-assigned detections, thus realizing multi-target tracking. The tracking method can be less affected by light changes and background noise interferences, thus solving the tracking failures caused by target blocking or mutual interferences between targets, providing multi-target tracking accuracy, and providing well instantaneity and robustness.

The invention discloses a detecting and recognition method of moveable and visible foreign matters and bubbles in liquid medicines, which is characterized in that the method comprises the following steps: (1) a continuous multi-frame image of a liquid medicine to be detected is obtained; (2) the image is preprocessed; (3) a moveable object is extracted, and a reinforced image is obtained; (4) the moveable object is partitioned as follows: an improved two-dimensional maximum entropy threshold partitioning algorithm is applied into the processing of the reinforced image and a two-dimensional histogram that is formed by grayscale and average neighborhood grayscale is adopted for choosing threshold, thus obtaining a partitioned image; (5) the movable object is traced; and (6) the image is identified and determined as follows: a movement trace of the object is obtained according to the tracing result, objects inducing foreign matters and bubbles are identified and distinguished by utilizing the continuity and the directivity of the object movement trace, and consequently whether the liquid medicine is qualified is finally determined. Under the condition of ensured algorithm precision, the detecting and recognition method of moveable and visible foreign matters and bubbles in liquid medicines adequately simplifies computation, improves the real-time property of the algorithm, greatly raises the accuracy of foreign matter and bubble identification in the liquid medicine, and lowers the false detection rate of the medicine.

The invention relates to a robust ear detection method which improves an AdaBoost arithmetic and belongs to the technical field of image mode identifying. The invention is characterized by proposing an ear detection method with excellent performances under a complex underground. The invention proposes four anisomerous Haar-like corner characteristics which are used for describing the grayscale changes on the partial areas of the ears; a policy of subsection selection is adopted for selecting the best sorting threshold of the Haar-like characteristics, thus reducing the sample training time; the weight of a weak sorter is modified for reducing the mistaken detection rate of the sorter; the threshold HW is set and eliminated according to the distribution change of the sample weight in the training, thereby preventing an over-studying phenomenon from being generated and leading the miss-detection rate and the mistaken direction rate of the ear detection to be reduced; besides, the invention also provides a single-ear detection policy for leading both the defection efficiency and the detection effect to be improved. The excellent performances of the robust ear detection method are shown on a PC machine and a DSP.

The invention discloses a target detection and recognition method based on deep learning. A model is built by utilizing a caffe platform in a Linux system, and consists of convolution layers, pooling layers, RPN layers, an ROIs layer, an ROI Pooling layer, fully-connected layers, a Sparse PCA layer and prediction window and prediction category output layers. A multi-RPN-layer fusion strategy is provided by the invention, and the detection capacity for a target of each scale is enhanced; the new Spares PCA layer is added between the two fully-connected layers, and the accuracy is guaranteed while the computational amount is reduced; and lastly, a logarithmic suppression method is provided for target location prediction, and a target location regression algorithm is improved. Finally, the purpose that the detection speed is accelerated while the detection precision is guaranteed is achieved. The method realizes accurately and quickly detecting and recognizing the targets of interest, and has a great application value.

The invention belongs to the field of computer security, and relates to a method for detecting web page Trojan horse based on program execution characteristics, which comprises the following steps: using web crawlers to capture source codes of a web page; then obtaining a recognizable script program through multilevel decoding; carrying out disassembling processing on the script program to obtainassembled source codes while reserving the script program; then, judging whether a large number of filled invalid instructions, calling system level functions and obvious URL links exist in the sourcecodes; and finally detecting whether the Trojan horse exists in the web page through the assembled source codes in a deep level. Because most of the prior web pages with the Trojan horse are embeddedwith ShellCode, to execute the ShellCode in the web pages in a local computer, system vulnerability is needed to realize buffer overflow and enable the program to skip onto the ShellCode code segment. Thus, only by analyzing the condition of executing the ShellCode, and analyzing the source codes according to the execution characteristics, whether the web page to be detected is the web page Trojan horse can be quickly detected.

The embodiment of the invention discloses a skin-tone detection method and a skin-tone detection device. The method comprises that: skin-tone pixels collected by a training specimen are clustered in a color space to obtain at least one skin-tone chroma class; a candidate skin-tone region in the training specimen is extracted, the distance between a chroma mean value of the candidate skin-tone region and the center of the skin-tone chroma class is computed, the training specimen is fallen under the skin-tone chroma class with the smallest distance to obtain a training subset corresponding to the skin-tone chroma class; the skin-tone probability distribution and the non skin-tone probability distribution of each training subset are statisticized to obtain a skin-tone probability model that each skin-tone chroma class corresponds to; an image to be detected is received, the candidate skin-tone region of the image to be detected is extracted, the distance between the chroma mean value of the candidate skin-tone region and the center of the skin-tone chroma class is computed, pixels in the image to be detected are made skin-tone discrimination according to the skin-tone probability model that the skin-tone chroma class with the smallest distance corresponds to, and the pixels discriminated as the skin-tone form a skin-tone region image. With the invention, the miss detection rate or the error detection rate of the skin-tone can be decreased.

The invention relates to a detection sorting device for steel balls The detection sorting device comprises a rack, a feed mechanism, an unfolding mechanism, a stock stop, a detecting mechanism, a sorting mechanism and a material collecting mechanism, wherein the feed mechanism comprises a material storing box, a material discharging hopper and a material discharging pipe; the unfolding mechanism comprises a material transporting guide rail, the material transporting guide rail is a V-shaped linear guide rail, the material discharging hopper is connected with an initiating end of the material transporting guide rail through the material discharging pipe, and the detecting mechanism comprises a light source assembly and a camera assembly; the light source assembly is arranged on the material transporting guide rail in a sleeved mode, the sorting mechanism and the material collecting mechanism are arranged on the tail end of the material transporting guide rail, and the separate collection of defect steel balls and qualified steel balls are realized through the sorting mechanism and the material collecting mechanism. The detection sorting device provided by the invention has the advantages that the automatic feed, detection and sorting of the steel balls are realized, a detection process is that two-dimensional motion of a traditional unfolding disc is changed into single one-dimensional rolling of one-sided steel balls, the detection efficiency is high, a false drop ratio and an omission ratio are low, and the steel balls are not magnetized or the surfaces of the steel balls are not damaged in a detection process.

The invention discloses a website weak password detecting method which comprises the steps of conducting automatic scanning on a website based on a dispatching task to acquire user information of the website, wherein the user information comprises the user name, the code and the encryption method of the code; encrypting a weak password in a weak password dictionary based on the encryption method to acquire an encrypted weak password; matching the encrypted weak password with the code to acquire a weak password code and a normal code; conducting simulated login verification based on the weak password code and the user name corresponding to the weak password code. According to the method, weak password detection is efficiently and accurately conducted on the website, the omission ratio is low, and pressure on the website to be monitored is low.

The invention discloses a method for detecting redundant code defects based on static analysis, for solving the problem lacking of a mature method for detecting redundant codes and related defects currently. The method comprises the following steps of: 1, inputting a program to be tested; analyzing the program into an abstract syntax tree; 2, analyzing statements possibly including operations of explicit power and the like; detecting the operations of the explicit power and the like; 3, for locally defined variables, detecting redundant assignment statements by using an internal process analysis method; 4, traversing the abstract syntax tree of the program; searching structures including the defects on the basis of a standardized program dependency graph; detecting dead codes; 5, detecting redundant conditional expressions; 6, detecting operations of implicit power and the like; and 7, detecting redundant function parameters; and giving a defect detection report according to the six defect detection results obtained in the steps 2-7. The method disclosed by the invention is applicable for analyzing large-scale program codes.

The invention provides a welding quality monitoring system and method. The welding quality monitoring system comprises a welding gun, a thermal infrared imager, an image capture card and an image processing computer. The welding gun is used for welding welding spots. The thermal infrared imager is used for recording molten bath infrared thermal imaging data of the welding spots. The image capture card is used for collecting the molten bath infrared thermal imaging data and sending the molten bath infrared thermal imaging data to the image processing computer. The image processing computer is used for analyzing the molten bath infrared thermal imaging data to obtain welding parameters, comparing the welding parameters with preset welding standards and obtaining a welding spot quality estimation result. In this way, by the adoption of the welding quality monitoring system and method, online monitoring of automatic welding quality can be achieved, the monitoring efficiency and monitoring quality are improved, and the missing inspection rate is reduced.

The invention discloses a method and equipment for the visual detection of visible foreign matters in a bottled liquid medicine. The method comprises the following steps of: (1) obtaining a multi-frame continuous image from a current liquid medicine which is to be detected and is filled in a medicine bottle by using an industrial camera; (2) preprocessing the image: carrying out top-hat morphologic filtering on the obtained image by adopting a 7*7 circular template as a structural element; (3) extracting a moving object; (4) tracking the moving object; and (5) identifying and judging the image. The device comprises a bottle grabbing manipulator used for grabbing the medicine bottle and an image collecting and processing device arranged close to the bottle grabbing manipulator. The method and the device are suitable for detecting the visible foreign matters of a traditional small-capacity bottled liquid medicine and have the advantages of simple realization, good detection effect, high safety and reliability, extensive application range and the like.

The invention discloses a traditional Chinese medicinal material microscopic image noise filtering system and method adopting a pulse coupling neural network. A pulse coupling neural network model suitable for processing tissue image information is adopted for detecting traditional Chinese medicinal material microscopic images. When small-density pulse noise pollution happens to the traditional Chinese medicinal material microscopic images, adaptive weighted filtering processing is carried out; when large-density pulse noise pollution happens to the traditional Chinese medicinal material microscopic images, the introduction dual-structure element mathematical morphology with edge detailed information kept is adopted for secondary filtering. Earlier foundations are laid for further improving quality testing, recognition and identification of traditional Chinese medicinal materials; higher noisy point detection performance is achieved, the noise fallout ratio and omission ratio are low, and the noise detection precision is high; detection time is short, and automatism is high; noise is removed, meanwhile, noise interference can be effectively filtered out, and image edge details and other information can be protected well.

The embodiment of the invention provides a flame monitoring method and system based on video camera. The method comprises the following steps of: collecting a video image; establishing a brightness Gaussian model by each collected frame video image and obtaining at least one high-light area and boundary pixels thereof from the image by utilizing the model; carrying out motion characteristic analysis on each collected frame video image by utilizing a Gaussian frame difference method to obtain movement foreground pixels; when the number of the movement foreground pixels in the boundary pixels in the high-light area reaches a pre-set threshold value, marking the high-light area to a flame suspected area; estimating blinking characteristics of the flame suspected area by utilizing an HMM (hidden Markov model) and judging whether the flame suspected area has the flame; when the flame exists, carrying out fire alarm; or continually monitoring the video image of the next frame.

The invention discloses a non-contact pipeline magnetic detection method which relates to a damage detection technology of petroleum and natural gas buried metal pipelines. The non-contact pipeline magnetic detection method comprises the following steps that (1) initial analysis is conducted and the construction period is determined; (2) the ground is cleared and a route is marked; (3) a pipeline is detected; (4) detection data is analyzed and processed; and (5) a data result is processed. According to the non-contact pipeline magnetic detection method, the defects of an existing contact method for detecting the local defects of the pipeline, such as a complicated process in which blind excavation is carried out, the pipeline is cleared and demagnetized, and finally the pipeline is uniformly magnetized, are overcome; and the pipeline is uniformly magnetized by the earth magnetic field, on-line detection can be carried out on the surface of a work piece without clearing the surface of the detected work piece or pre-treating the surface of the detected work piece in other ways, so that the detection process is greatly simplified, the speed is high, the efficiency is high, and the detection depth is greatly increased.

The invention relates to an LCD defect detection method based on a feature pyramid convolutional neural network. The method comprises the steps of constructing a feature pyramid convolutional neural network and constructing a detection model, detecting an LCD panel picture by using the model; defect types and positions are screened and determined through output parameters. According to the method,respective advantages of the deep residual convolutional neural network and the feature pyramid network are fully utilized. According to the invention, the low-layer high-resolution image features and the high-layer high-semantic information features are effectively fused, and the type and position of the defect are directly obtained from the LCD picture by adopting single-stage detection, so that the detection efficiency is greatly improved, and the omission ratio and the false detection rate are reduced.

The invention belongs to the technical field of image processing and pattern recognition, and relates to a cloth defect detecting method based on machine vision. The cloth defect detecting method includes performing power spectral density analysis for an image of a normal cloth texture and acquiring central frequency F and an azimuthal angle theta of the texture; constructing an SXL adaptive Gabor filter bank; filtering the image of the normal cloth texture to obtain a feature image group, and computing the mean value and the variance of each image of the feature image group; acquiring an image of to-be-detected cloth; filtering the image of the to-be-detected cloth to obtain a feature image group; performing threshold post-processing for the feature image group of the image of the to-be-detected cloth to obtain an absolute feature image group; carrying out normalization processing; fusing images and performing binarization processing for the images to obtain a detected binary image; and removing noise interference to obtain a final detection result. The S represents the number of the selected central frequency, and the L represents the number of the selected azimuthal angle. The cloth defect detecting method has the advantages of high universality and efficiency.

The invention relates to a pedestrian navigation system three-dimensional spatial positioning method based on human / environment constraints and solves the problem with high-precision positioning of a pedestrian navigation system in a three-dimensional space, wherein the system is designed based on a low-cost MEMS (micro-electromechanical system) inertial measurement unit; data collection and strap-down navigation calculation are performed by installing the MEMS inertial measurement unit on a shoe, and zero-speed ranges in pedestrian gaits are detected in connection with data output by a pressure sensor installed on a sole. A scheme is provided based on zero speed correction in connection with pedestrian motion characteristics and ambient building characteristics, the scheme is about constraining magnetic field abnormality in zero-speed ranges and constraining SINS (ship's inertial navigation system) height channel divergence, and an extended Kalman filter is designed to correct errors in navigation calculation results, thereby enabling high-precision positioning of pedestrians in the three-dimensional space.

The invention discloses a remote sensing image airplane detection method based on fusion of angle points and edge information. The method comprises the following steps of: executing edge detection on a remote sensing image; performing binarization on the detected remote sensing image; executing Harris angle point detection on the binarized remote sensing image; selecting regions in which the number of the angle points is in a predetermined range as candidate airplane regions; removing false target regions according to the number of bright pixels in each candidate airplane region; clustering the obtained image and marking a target airplane position, thus finally obtaining the number of obtained clusters and a central position of each cluster, wherein the number of the obtained clusters is the number of detected airplanes, and the central position of each cluster is the central position of a target airplane. The invention further discloses a remote sensing image airplane detection system comprising an edge connection unit, a binarization processing unit, an angle point detection unit, a candidate airplane region selection unit and a clustering unit which are orderly connected. Compared with the prior art, the method has the advantages of better detection effect and higher detection efficiency.

The invention provides a primer set for simultaneously detecting mutation of a plurality of regions of two or more exons of one or more cancer drive genes in a sample on the basis of multiplex PCR (polymerase chain reaction) and high-flux sequencing techniques, a kit comprising the primer set, and a method for amplifying the cancer drive gene or detecting the cancer drive gene by using the primer set and / or kit. The mutation detection result can be used for instructing cancer clinic medication, auxiliary diagnosis or prognosis judgment on certain cancers.