111results about How to "Improve target tracking accuracy" patented technology

The invention discloses a target tracking method with multi-spectrum image sensor fusion. The method adopts a four-level data fusion computing framework with certain versatility. The four-level data fusion computing framework is respectively a multi-characteristic fusion level, a multi-mode fusion level, a multi-model fusion level and a multi-sensor fusion level from top to bottom. The multi-characteristic fusion realizes the integration of a plurality of characteristic clues through the probabilistic method to guarantee enough characteristic information representing targets in different environments; the multi-mode fusion adopts handoff algorithm in a staging mode, integrates two adjacent tracking modes through a certain weighting coefficient, to ensure smooth transition between the modes and guarantee continuous tracking of the targets; the multi-model fusion adopts an interacting multi-model algorithm to solve the problem of stable switching among different moving models of the targets; the multi-sensor fusion adopts distributed type fusion tracking algorithm to select the advantages of all the sensors, and makes up the deficiencies between the sensors, to ensure the tracking to be more robust and more precise.

The invention relates to a deep learning-based target tracking method, a deep learning-based target tracking device and a storage medium. The method includes the following steps that: two frames of pictures are continuously read; a target region of the former frame and a search region of the current frame are set and cut, when the search region of the current frame is set and cut, whether an object is stable when the object moves fast is judged, so that a center point position is set, and the search region can be obtained; the target region and the search region are inputted to a convolutional neural network, calculation is performed, so that the target region of the current frame can be obtained; calculation is performed, the inter-frame displacement of the current frame relative to the former frame is obtained; and whether the current frame is the final frame is judged, so that whether iterative target tracking is performed further is judged. According to the deep learning-based target tracking method, the deep learning-based target tracking device and the storage medium of the present invention, the prediction of the center point position of the cutting region of the current frame can be realized through judging the speed of the movement of the target object in the image. Compared with an existing algorithm, the method can improve target tracking accuracy and target coincidence degree with original high tracking speed maintained, and has good tracking robustness.

An early stage tumor localizing tracking based on the multimode sensitization imaging fuse, belongs to the medical image processing field. The invention includes: a medical image before the operation for obtaining the tumour aim focus imaging sensitization; an ultrasound sensitization image during the operation for obtaining the tumour aim focus imaging sensitization. When the image is processed the guide therapy, using the global rigid transformation and the local nonstiff transformation combination round the tumour aim focus as the geometric transformation model with deformation registration, the sensitization images before and during the operation are processed with the deformation registration based on the union marked region, while the images before and during the operation are fused, to rebuild the three-dimensional visualization model in the tumor focus region. Using the above deformation registration method to complete the sport deformation compensation for the imaged before the operation, the target tracking of the tumour target focus is further automatically completed. The invention can be used in a plurality of places, such as the early diagnosis of the tumour, the image guide tumour early intervention, the image guide minimal invasive operation, the image guide physiotherapy etc.

The invention belongs to the technical field of radar moving target tracking and particularly relates to a before-radar-target-detecting tracking method based on track enhancing. According to the tracking process, first, for multi-frame original data, non-parametric track initial batch processing operation is carried out, and a target trace point set is obtained; then according to target track state parameters updated in real time and errors in measurement, a corresponding parameterization track enhancing operator is designed; and finally current measuration is combined, and a parameterization track enhancing method is used for carrying out prediction tracking. The method is characterized in that closed loop feedback is achieved, during track enhancing operation, multi-frame measuration and states can be fully used, detecting tracking accuracy and precision can be improved, the method has the advantage that under the condition of the low signal-to-noise ratio, detecting and tracking integration processing can be achieved, and the method can be used in the field where monitoring systems such as radar are used for detecting and tracking a long-distance highly-maneuvering target with weak echo signals.

The invention belongs to the technical field of intelligent automobiles, specifically relates to a complex environment radar multi-target tracking and road driving environment prediction method, whichparticularly aims to solve the problems that the position relationship identification of a lane where a target vehicle locates is inaccurate and the robustness and precision of a target tracking algorithm are not high through determination based on a radar original target measurement value in the process that a vehicle having a self-adaptive control function drives in a curve or an intelligent vehicle having an autonomous valet parking function enters and exits from a curved ramp of an underground parking lot. The complex environment radar multi-target tracking and road driving environment prediction method is mainly implemented by present vehicle motion state estimation, millimeter wave radar signal conversion, time synchronization, target motion compensation, data rationality judgment,target measurement value noise reduction, road curvature estimation, target aggregation, target motion attribute and motion state identification, improved adaptive extended Kalman filtering algorithmtracking and data association, road driving environment prediction and key target generation.

The invention relates to a distributed MIMO radar target tracking accuracy joint resource optimization method. The method includes the following steps that: an objective function is derived, a resource allocation optimization model containing three variables, namely, array element, power and bandwidth is constructed; the model is simplified, an array element selection problem is converted into a convex optimization problem from a 0-1 problem through convex relaxation; and the optimization model is decomposed into iterative optimization sub-problems of array element selection and power allocation through using cyclic minimization, the problems are solved through using an SPCA (sequential parametric convex approximation) method until the objective function is not optimized any more, final array element selection and power allocation results are obtained, and a final bandwidth allocation result is calculated, and a target position at a next time point is estimated. According to the distributed MIMO radar target tracking accuracy joint resource optimization method of the invention, the accuracy of the target tracking of an MIMO radar can be effectively improved through resource allocation, and the utilization rate of resources can be also improved; and an appropriate array element subset is selected, and power and bandwidth resources are allocated optimally, and therefore, the accuracy of the target tracking can be further improved under a situation that radar system resources are limited.

The t invention discloses an improved multi-expansion target tracking method. The method comprises the following steps: carrying out clustering and division on a measurement data set of a sensor and initializing a probability hypothesis density value and a potential distribution value of a target; predicting and updating a probability hypothesis density value and a potential distribution value oftarget state set at a next time to obtain a probability hypothesis density value and a potential distribution value at the current time; carrying out trimming and combination on a Gaussian term of anintensity function of the target, extracting target state estimation, and carrying out performance evaluation; and carrying out prediction, updating, and trimming and combination repeatedly and tracking the target till disappearance of the target. Therefore, the tracking accuracy of the target in a clutter environment is improved; the precision losses caused by the blind area of the radar are reduced; the computing load of the filter is reduced; and the engineering application of the GM-CPHD filter is realized well.

The invention provides a detection system for detecting targets around a vehicle and application of the detection system. The detection system includes a vehicle front forward direction sensor (1), a vehicle roof forward direction sensor (2), a right front sensor (3), a left front sensor (4), a right rear sensor (5), a left rear sensor (6), a rear sensor (7), a left side sensor (8), a right side sensor (9) and target selection modules (10), wherein the target selection modules (10) include a forward direction target selection module (11) and a side direction and backward direction target selection module (12), wherein the forward direction target selection module is configured to select targets at least four positions in front of the vehicle as vehicle front side tracking targets based on received target motion information, and the side direction and backward direction target selection module is configured to select targets at the side direction and backward direction of the vehicle as vehicle side direction and backward direction tracking targets according to received target motion information.

The invention belongs to the technical field of photoelectric tracking, precision measurement, navigation and location and particularly relates to a target tracking method based on intelligent IMM Kalman filter feedforward compensation and a system. A neural network intelligent target recognizer recognizes the type of a tracked target, state equation parameters of online operation of an improved intelligent IMM Kalman filter are adjusted, so that the filter is endowed with an adaptive characteristic, an online operating state model equation is matched with the tracked target, multiple filtersbased on different models are used for parallel work, speed and acceleration of the maneuvering target are estimated in real time accurately, estimation accuracy higher than that of traditional filtering is obtained, and a feedforward compensation controller constructed on the basis can obtain higher feedforward compensation precision and control performance.

The invention discloses a target positioning and tracking method for a wireless sensor network. The method comprises the following steps: at any positioning time, estimating a position of a target according to measurement information of a sensor node, building a learning region covering the estimated position of the target, selecting an arbitrary quantity of position points in the learning region, obtaining a decision function by use of a polynomial kernel function and a mapping relationship of the distance vector from an approaching position point of an epsilon-support vector machine for regression to the sensor node and a coordinate of the position point; and obtaining an estimated value of the position of the target by inputting a distance vector from the sensor node to the target into the decision function and transmitting an estimated value of the position of the target to a base station which then fits the historical data of the position of the target and upgrades the motion patch of the target to complete target tracking. The method obviously reduces the influence of measurement errors of the sensor node on the target positioning and tracking and improves the accuracy of target tracking.

Provided is an infrared / laser radar data fusion target tracking method based on a multi-scale model. The method comprises the steps that the multi-scale model is built; data collection is carried out on a target, and the distance measurement value, the azimuth angle measurement value and the pitch angle measurement value of the target are obtained; estimation is carried out on angle information obtained by an infrared detection system on a scale 1 through the unscented Kalman filter method; the angle estimation information and the distance information are fused on a scale 2; the target state is estimated through the Kalman filter method on the scale 2; the fused and estimated information is converted on the scale 1, filtering estimation is carried out on the scale 1 through the Kalman filter method, and an accurate filtering estimated value is obtained; the above steps are repeated, and a target movement track is obtained. The multi-scale model is quoted into information fusion, the target movement state is described more comprehensively and accurately, estimation filtering is carried out on the target on different scales, the estimation precision of the target state is improved, and the target tracking precision is improved through information interaction between different scales.

The present invention provides a multi-target tracking method after through-wall radar imaging, the rapid imaging can be achieved in a certain error, and a high quality imaging result can be obtained. According to the target targeting algorithm based on two-dimensional low-pass filtering and image etching, measurement position information and amplitude information can be effectively extracted, and finally a nearest area data association algorithm is used to carry out target tracking. At the same time, according to an interacting multiple model tracking algorithm, the high mobility of the target is fully considered, and the target tracking accuracy is high.

The invention discloses an improved multi-extended target tracking method, which comprises the steps of 1, clustering; 2, initialization; 3, predicting and updating; 4, pruning and merging; and 5, repeating the step 3 and the step 4. In allusion to a multi-extended target tracking problem in a clutter environment, measurement data of extended targets is effectively processed through adopting a DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, and effective tracking for the multiple extended targets is realized. Due to the introduction of a scale factor and the design of an adaptive threshold, the precision loss caused by a blind area of radar is reduced, the calculation amount of a filter is reduced, and the engineering application of a GM-CPHD filter is facilitated.

A satellite laser distance measurement system based on tilt calibration is featured as setting high speed tilt reflector in telescoping unit, using tilt tracking transducer to carry out wave front detection on light wave received by receiving telescope, outputting calibration control signal to high speed tilt mirror in telescope by high speed processor after error signal is restored, starting up satellite laser distance measurement system in telescoping unit to send out laser with specific wavelength for carrying out satellite laser distance measurement after tilt closed loop real time calibration is finished.

The invention belongs to the technical field of image processing, and relates to a multi-source target tracking measurement system and tracking method based on active vision. The system comprises a photoelectric assembly and a stable platform, wherein the photoelectric assembly is composed of a visible light camera, an infrared thermal imager, a laser range finder, an information processor and a photoelectric cabin structure; the visible light camera, the infrared thermal imager, the laser range finder and the information processor are all installed inside the photoelectric cabin structure; the stable platform consists of an orientation motion assembly, a pitching motion assembly and a control assembly; and the control assembly receives an instruction sent by the information processor, andcontrols the orientation motion assembly and the pitching motion assembly to enable visual axes to stably point to a target. The invention further provides the multi-source target tracking method based on the active vision. The orientation motion assembly and the pitching motion assembly are driven to rotate according to the control instruction, so that the visual axis of the visible light cameraand the visual axis of the infrared thermal imager point to the target, and the target is always kept in the center of a shot image.

The invention discloses a track correlation method based on a LHD gray correlation degree. The method comprises acquiring multiple target information scanned by multiple sensors, carrying out processing to obtain a target track sequence, computing a grey relational degree between target tracks acquired by multiple sensors, calculating a gray correlation matrix according to the grey relational degree, determining a grey relational threshold to obtain a track correlation matrix and determining track correlation according to the track correlation matrix. The track correlation method has a high target track precision, and fast and correctly associates with the target track in a dense cluttered environment. The method effectively improves the accuracy of the track association algorithm, has a small amount of calculation and can be well used in the engineering practice.

The invention belongs to the technical field of computer vision, and particularly relates to a target tracking method based on a Siamese FC framework and a PFP neural network, which comprises the following steps: (1) based on the Siamese FC framework, processing a target area in a first frame of a video to obtain template characteristics; Inputting the template features into a PFP neural network to obtain template final features; (2) based on a Siamese FC framework, performing t-detection on the t- Processing the target area in one frame to obtain a search area feature; Inputting the featuresof the search area into a PFP neural network to obtain final features of the search area; (3) taking the final feature of the template as a convolution kernel, performing convolution on the final feature of the search area, and determining a central position of a tracking target in the t-th frame and a target area; And (4) repeating the step (2) and the step (3) until the video is finished, and completing the tracking of the central position of the tracking target and the target area. According to the invention, the context information and different receptive field information can be fused, sothat the target tracking precision is improved.

A method and system is proposed for use by a moving station (such as a jetfighter) for radar tracking of a moving target (such as an air-to-air missile). The proposed method and system involves the use of a hybrid FSK / LFM (Frequency Shift Keying & Linear Frequency Modulation) scheme for acquiring a collection of raw radar data, a first Gaussian-noise filter array of one-stage linear Kalman filters for S / N-enhancement of the raw radar data, a trilateration module, and a second Gaussian-noise filter array of one-stage linear Kalman filters for S / N-enhancement of the trilateration-resulted radar data. These features allow the radar tracking of moving objects to be more fast and accurate.

The invention discloses an underwater target cooperative tracking method based on consistency estimation and dormancy scheduling. The method comprises the following steps: disposing two types of underwater sensor nodes, and arranging that the nodes are at a dormancy state with a low-frequency reflection detection function at initial time; the dormancy sensor nodes periodically determining whether to be converted into an activation state; sensors entering a wakeup state sending water sound pulse signals for detecting a target and accordingly determine the position of the target; for the purpose of obtaining target position information with quite high precision, performing consistency estimation on observation information of the two kinds of nodes; and according to a predicted target motion locus, dynamically activating the sensors disposed at a target prediction path area, setting work cycle periods and duty ratios for the nodes, and at the same time, underwater dynamic sensors dynamically adjusting positions so as to realize continuous and dynamic tracking of the target. The method provided by the invention has the advantages of improving the target tracking precision, reducing the energy consumption, prolonging the network life and the like.

The invention discloses a monitoring target linkage tracking control method and device and a storage device. The method comprises the following steps: acquiring an initial image and initial coordinates of a monitoring target; controlling a first monitoring camera device and a second monitoring camera device to track and shoot the monitoring target according to the initial coordinates; acquiring afirst tracking image of the monitoring target tracked and shot by the first monitoring camera device, and a second tracking image of the monitoring target tracked and shot by the second monitoring camera device; determining a first similarity value between a first monitoring target in the first tracking image and the initial image of the monitoring target, and determining a second similarity valuebetween a second monitoring target in the second tracking image and the initial image of the monitoring target; and adjusting the first monitoring camera device and the second monitoring camera device according to a similarity comparison result. In this way, the purposes of improving the target tracking precision and preventing the target from being lost can be achieved.

The invention provides a radar node selection method based on measurement error covariance matrix norm. The radar node selection method comprises the following steps: radars for constructing the least squares norm of a measurement error covariance matrix is taken as target tracking operating radars, and at last, performing target tracking process by the selected radars. The least squares norm of the measurement error covariance matrix is used as a criterion for selecting radar nodes and the correlations between different radar measurement values is fully considered. Under the condition that the correlations between different values are fully considered, radars which are small in measurement error, more in effective information and less in redundant information are selected out to track targets, so that the target tracking precision is improved.

The invention relates to a joint resource optimization method for multi-target position estimation in a distributed MIMO radar system. The joint resource optimization method comprises: a target is designated and a minimized maximum value of a multi-target position estimation error is used as a target function; under the circumstances that the total numbers of the transmitting and receiving array elements are limited and the transmitted power is given, a resource optimization model of combination of transmitting and receiving array element selection and power allocation is established; and on the basis of a heuristic search algorithm and a continuous parameter convex approximation algorithm, a resource joint allocation algorithm based on cycle minimization is proposed to solve a hybrid Boolean type joint optimization problem to obtain a joint resource allocation result. According to the invention, the energy relationship between system resources and tracking capabilities is analyzed quantitatively; compared with the array element number, the influence on the system performance by the transmitted power becomes obvious and the influence on the target tracking precision and number by the system resources is displayed, so that the computing load of the system is reduced, the good system performance is realized, and the overall multi-target tracking accuracy is improved. The joint resource optimization method has the great practical application value.

The invention discloses a CS-Jerk model-based parameter adaptive maneuvering target tracking algorithm. According to the algorithm, on the basis of a CS-Jerk model and current statistics thoughts, truncation probability distribution is utilized to describe current probability density of the acceleration change rate of a target, so that the relation of the variance of the acceleration change rate of the target and the Jerk mean is obtained, and therefore, the adaptive adjustment of the variance of the acceleration change rate of the target is realized; the change of the maneuvering condition of the target is judged through using residual vectors; and the adaptive adjustment of maneuvering frequency is realized through a nonlinear maneuvering frequency function; and the adaptive adjustment of a process covariance matrix Q (k) is realized, and the problem that a CS-Jerk model needs the manual setting of a process covariance matrix can be solved, and target tracking performance can be improved.

The invention discloses a multi-camera multi-target tracking method. The multi-camera multi-target tracking method comprises the steps of collecting an initial track fragment set collected by multiplecameras; carrying out equalization processing on the trajectory fragment similarity distribution of the initial trajectory fragment set to obtain a pre-processed trajectory fragment set; and based onthe pre-processed track segment set, fusing each pre-processed track segment to obtain a complete track of each target, and completing the multi-camera multi-target tracking. According to the invention, by adopting a one-step walking idea, the track segments under the plurality of cameras are put together for track fusion; moreover, when all the tracks are fused, the equalization processing is carried out on the similarity distribution of all the track segments, so that the similarity of all the track segments corresponding to the single target is ensured to be at the same level, the efficiency and the effectiveness of multi-camera multi-target tracking are improved, and the problem that the accumulated errors exist when the single camera track segment fusion is firstly carried out and then the cross-camera track fusion is carried out, is solved to a certain extent.

The invention provides a slow moving target elimination method based on a point clutter map. The method is used for improving tracking precision of oversea low-altitude flight target, and comprises the following implementation steps: calculating a Doppler-distance dimension data matrix Dj on an initial orientation alpha received by a radar in each frame j; calculating a test statistics matrix lambda j of the Doppler-distance dimension data matrix Dj; performing Doppler condensation on the test statistics matrix lambda j; acquiring an initial test result Hj; extracting a slow Doppler channel index j set M'; obtaining the test statistics matrix (represented as a formula in the specification) of a slow Doppler channel; getting rid of the slow moving target in the test statistics matrix (represented as a formula in the specification) of each frame j; and getting rid of the slow moving targets in pulse-distance dimension data in the other orientations, except for the orientation alpha, in all the frames. With the method provided by the invention, performance of eliminating the slow moving target in a low speed channel is improved, and the method can be used for elimination of sea surface slow moving targets uninterested under a radar on sea mode.

The invention discloses a real-time target tracking method based on online study. Through structuring a video collecting module, an ELM classifier initial training module, a target tracking module, a classifier updating module and a video storage module, the real-time target tracking method based on online study is realized. The video collecting module reads image data, the ELM classifier initial training module forms an initial ELM classifier, and the target tracking module tracks and positions the target; the ELM classifier updating module updates the ELM classifier and the video storage module to compress and encode the video. A target location based on a particle filter uses limited particle number to generate candidate samples, thus the calculation amount is reduced, and the calculating efficiency is improved; the ELM classifier based on the online study realizes the real-time gain study of the training sample, rapidly adapts to the target appearance change, and promotes the target tracking precision.

The invention discloses a real-time tracking method for an infrared image target with multi-feature fusion. The real-time tracking method comprises the following steps of: initializing a target tracking point position; initializing a target model; calculating a target candidate model; calculating a coefficient of unified feature Bhattacharyya and weight coefficients between feature; calculating a new tracking position of a current frame target; estimating the coefficient of the unified feature Bhattacharyya at the new position; and comparing the two coefficients of the unified feature Bhattacharyya and outputting results. The real-time tracking method can be used for adaptively calculating the weight coefficients between multiple features, enhances the robustness of target tracking, ensures the stability of target tracking, solves the problem of tracking point drifting caused by unstable single feature, and effectively improves the accuracy of target tracking.

The invention relates to a quick real-time discrimination type tracing method based on multi-local-feature learning. The method comprises the following steps of: single-frame decomposition is carried out on a video, a target to be traced is marked in an initial frame through combination with early-stage action identification work or an artificial marking method, then, candidate blocks are obtained in a candidate area through dense sampling, and the local feature of each candidate block is independently calculated; then, on the basis of a circulant matrix and a relevant filter, a classifier is trained in a Fourier domain, a detection formula is used for calculating correlation between each candidate block in a current frame and the target also in the Fourier domain in a position detection stage, and an area with the highest correlation is selected as a prediction position of the target of the current frame; and finally, in the prediction position, a position detection result which has a smallest difference with a previous frame of target is selected as a final tracing result of the current frame, and a new target feature is used for training a new classifier. Except the initial frame, the tracing of other frames is characterized in that position detection is firstly carried out, and then, the training of a new regression function is carried out.

The invention belongs to the field of radar target tracking, and particularly relates to a spatial-temporal resource-waveform selection management method for a multi-beam centralized MIMO radar. The spatial-temporal resource-waveform selection management method comprehensively considers target detection and target tracking, and achieves efficient allocation of system resources in tracking processwhile improving target tracking accuracy. When solving a related algorithm, the proposed algorithm first selects combinations of a feasible sub-array division, a sampling periods, emission energy, anirradiated target set and sub-array beam pointing set parameters through a target successful irradiation limit and a target effective detection probability limit to adaptively allocate radar system resources; then further improves tracking accuracy, combines transmitted waveform parameters to form combinations of the feasible sub-array division, the sampling period, the emission energy, the irradiated target set and the sub-array beam pointing set parameters and the transmitted waveform parameters, and finally selects an optimal combination of the sub-array division, the sampling period, the irradiated target set and the sub-array beam pointing set parameters and the transmitted waveform parameters according to principle of target function minimization.