135 results about "Scale estimation" patented technology

The invention discloses a scale adaptive target-tracking method based on a depth characteristic kernel correlation filter. The method comprises the following steps: inputting an image into a pre-trained convolution neural network, and extracting depth convolution features; tracking a target, and estimating the position and scale of the target through the trained model; training the kernel correlation filter according to the currently detected target position and scale; and updating the kernel correlation filter through employing an adaptive high-confidence model updating method. According to the invention, the depth convolution features are extracted, and the adaptive scale estimation method and the adaptive high-confidence model updating strategy are improved, thereby improving the target tracking robustness under the conditions of complex scenes and appearance changes. The method can achieve the high-efficiency and accurate processing of the scale change of the target. In addition, the adaptive high-confidence model updating strategy is employed, so the model tracking drift is reduced as much as possible.

The invention provides a kernel correlation filtering target tracking method based on feature fusion and Bayesian classification. The method is characterized by comprising steps: firstly, the position and the scale information of an initial frame target are given; then, a standard target appearance model, a standard correlation filter and a standard color Bayesian model are built or updated; based on the target center point of a former frame, a search area is extracted; and a correlation filter of a Gaussian kernel is used for realizing target displacement estimation, the color Bayesian model is used for realizing target scale estimation, and the tracking result of the current frame is obtained further. Through sequentially processing each frame of video image, tracking of a moving target in the video is realized. The target tracking method can effectively solve the problem of precisely positioning the moving target in the video and can realize target scale estimation. In multiple challenging environments, the target can be tracked effectively, certain robustness is realized, and the tracking algorithm precision is improved.

A condition-maintenance-mode city rail vehicle auxiliary maintenance system is characterized in that a city rail vehicle condition monitoring module is used for receiving and synthesizing data related to operation and maintenance, performing data screening and calculation on the basis of the data related to maintenance to form key data related to maintenance, and receiving and storing diagnosing and potential safety hazard data provided by a vehicular diagnosing mainframe and a city rail vehicle potential safety hazard mining server; a safety evaluation module is used for performing city rail vehicle safety evaluation on the basis of the condition monitoring data, and performing fault analysis, potential safety hazard information screening and service life evaluation aiming at the key parts of a city rail vehicle at the same time; a maintenance redundancy design module performs maintenance scale estimation and maintenance cost estimation aiming at the evaluation results and provides reports, operation and maintenance staff makes maintenance decisions, and a resource optimization allocation module provides system urgency and time estimation at the moment; a maintenance plan making module provides interfaces allowing the operation and maintenance staff to perform fine adjustment and makes maintenance plans for vehicle maintenance according to selections.

The invention relates to a scale adaptive kernel correlation filter fast target tracking method. According to the scale adaptive kernel correlation filter fast target tracking method, separation of afront background is promoted and a scale estimation module is added through introducing a two-dimensional variable bandwidth Gaussian window into a KCF framework, scale variations of a target are estimated according to a matching result of feature points of an image block, and whether the target is sheltered is judged by combining displacement prediction based on the scale variations. Once the target is sheltered, the user considers that no scale conversion occurs at the moment and template update is not performed, further the tracking precision is enhanced in this way. The scale adaptive kernel correlation filter fast target tracking method has the advantages of simple and fast calculation, and can significantly improve the calculation efficiency on the basis of compatible scale adaptation.

The invention provides a real-time visual object tracking method having shielding sensing mechanism, tracking results based on filter tracking algorithm based on correlation and Bayesian probability model tracking algorithm based on color statistical information are fused, position prediction of a target is performed, one-dimensional scale pyramid is independently trained to perform target scale estimation, in addition, a shielding sensing mechanism is designed, the mechanism can judge whether a target is partly or completely shielded, then an update strategy of the model and which kind of model utilized to track new frame are determined, tracker drifting caused by pollution of background information is avoided, and according to the above steps, the tracking algorithm is used for tracking the target persistently and iteratively. The above algorithm can be used for accurately tracking the target in real time, and a plurality of problems including deformation, rotation, shielding, size variation, illumination variation, motion blur and the like can be effectively handled.

The invention discloses a self-adaptive feature fusion-based multi-scale correlation filtering visual tracking method. The method comprises the following steps: firstly, the correlation filtering is carried out on a target HOG feature and a target color feature respectively by using a context-aware correlation filtering framework; the response values under the two features are normalized; weightsare distributed according to the proportion of the response values and then are subjected to linear weighted fusion, so that a final response graph after fusion is obtained; the final response graph is compared with a pre-defined response threshold value to judge whether the filtering model is updated or not; finally, a scale correlation filter is introduced in the tracking process, so that the scale adaptability of the algorithm is improved. The method can be used for tracking various features. The performance advantages of the features are brought into play, and a model self-adaptive updating method is designed. In addition, a precise scale estimation mechanism is further introduced. According to the invention, the updating quality and the tracking precision of the model can be effectively improved, and the model can be changed in scale. The method is good in robustness under complex scenes such as rapid movement, deformation, shielding and the like.

The invention provides a target tracking method based on a multi-resolution neural network. The method is suitable for use in video target tracking, and includes: receiving a to-be-detected current video frame, and extracting a detection block on the current video frame, extracting multiple layers of depth features which have different resolution and are of the detection block; utilizing a kernel correlation filtering model to calculate target displacement response graphs of all the layers of depth features; utilizing the multi-resolution neural network to carry out displacement estimation according to the target displacement response graphs of all the layers to obtain estimated target displacement, and updating the kernel correlation filtering model; and utilizing a scale estimation model to carry out scale estimation according to the estimated target displacement to obtain estimated values, and updating the scale estimation model. Through the embodiment, situations of tracking drift and tracking loss can be substantially reduced, and the robustness and the stability in complex environments of motion mutation, illumination variation and the like or long-sequence video tracking are substantially improved.

The invention discloses a multi-acquisition-instrument fingerprint crossing-matching method based on size scaling estimation. The method comprises the following steps that: A, a direction field of fingerprint is roughly calculated; direction values in a broken region are corrected; and the fingerprint according to the corrected direction field is enhanced; B, a method from thickness to thinness is adopted to estimate a size scaling value between template fingerprint and input fingerprint; C, a registration method based on detail-point clustering is adopted to determine candidate values of registration parameters, and the corresponding relation between detail-point sequences of the input fingerprint and the template fingerprint is established; and D, on the basis of all corresponding detail points, to similarity fractions for calculating normalization, multi-acquisition-instrument fingerprint crossing-matching based on size scaling estimation is realized. The method has the advantages of effectively and robustly processing the crossing-matching of multi-acquisition-instrument fingerprints, improving the interoperability of a matching method and realizing the robust matching of the multi-acquisition-instrument fingerprints, can be applied to the system with the coexistence of a plurality of acquisition instruments, and has important application value in the field of fingerprint recognition.

The invention provides a visual SLAM front-end pose estimation method based on deep learning. The visual SLAM front-end pose estimation method is used for estimating inter-frame pose transformation inreal time. The method comprises the following steps: firstly, carrying out data preprocessing on an original data set, and then constructing a Brox network to carry out dense optical flow extractionon input continuous frame images; carrying out feature extraction on the extracted optical flow graph by two networks, extracting high-dimensional features from one branch by adopting global information, dividing the optical flow graph into four sub-images at the other branch, and respectively carrying out down-sampling to obtain image features; and finally, fusing the features obtained by training the two branches, and carrying out pose estimation on the final cascade full-connection network to obtain a pose between two adjacent frames. According to the method, the problem of real scale estimation in monocular vision is solved, camera motion and proportion information can be extracted by using global information and local information, and the learning ability and intelligent level of therobot are improved.

The invention relates to the field of cross fusion in the field of computer vision and artificial intelligence, in particular to a pose estimation method based on self-supervised learning. The methodcomprises the following steps: S1, acquiring an image; s2, performing image preprocessing; s3, sending the current frame of image and the previous frame of image to a pose estimation network together,and solving pose estimation between the two frames, wherein the pose estimation network is obtained by training in a self-supervised learning mode by combining with a deep estimation network; and S4,solving the global pose of the current frame according to the global pose of the previous frame and the pose estimation between the current frame and the previous frame. According to the method, self-supervised learning is realized in a pose estimation network and depth estimation network joint training mode, an attention mechanism is added into the pose estimation network and the depth estimation network, finally, consistency scale estimation of long-sequence inter-frame pose transformation is realized, and the generated complete track is higher in precision.

The invention relates to a neutrosophic similarity measurement-based scale-adaptive visual target tracking method. The method comprises the following steps of: selecting a to-be-tracked target area inan initial frame and calculating a target feature histogram and an initial background histogram; carrying out truth, falsity and indeterminacy measurement aiming at target feature attributes and background feature similarity attributes; establishing a neutrosophic weight vector; introducing the neutrosophic weight vector into a mean shift strategy to determine a target area of a current frame; calculating corresponding truth, falsity and indeterminacy measuring values aiming at scale reducing and expanding and determining a scale updating strategy according to cosine similarity measurement; and updating a target background feature histogram. The method disclosed by the invention has the beneficial effects that an extremely efficient mean shift algorithm is adopted, the corresponding neutrosophic measurement calculating amount is small, the weight vector and scale estimating is low in complexity and high in efficiency and the requirements of real-time target tracking are met; and by utilizing a neutrosophic set theory, the tracking performance of a tracking algorithm coping with challenges of complex backgrounds and like is effectively improved through taking the change of trackedtarget features and the similarity of target / background features into account.

The invention discloses a scale estimation-based human face detection method. According to the method disclosed in the invention, a size of a human face on an image is estimated via human face scale estimation, the image is zoomed based on a human face scale, a suggestion frame extraction is performed rapidly via a full convolution network, and a human face detection result can be obtained via twotimes of cascaded classification and regression. According to the method, human face detection is performed based on combination of human face scale estimation and a cascaded convolution neural network, an overall calculation amount of human face detection can be reduced, overall time consumption for human face detection can be reduced, and human face detection effects can be ensured.

The invention belongs to the technical field of visual target tracking, and discloses a rotor operation flying robot target tracking method and system, and the method comprises the steps: taking a Siamfc framework as the basis, introducing Resnet50 as a feature extraction network through offset learning, enabling the network to learn more semantic information, and coping with the appearance changeof a target; according to the tracking network, a target scale estimation module is newly added on the basis of a classification discriminator, an IOU of a target bounding box and a target real box can be predicted, the target bounding box is accurately predicted, iterative correction is carried out on the bounding box through reverse gradient, and thus the network can accurately predict the scale change of a target; output of different layers of the network is fused by utilizing Resnet50 multi-layer feature output and adopting a residual fusion strategy, so that the robustness of the algorithm is further improved, the network performance is improved, the discrimination capability of the network to a small target is guaranteed, and finally, accurate tracking of the target is realized.

The invention relates to a high-speed correlation filtering tracking method based on a high-confidence updating strategy. A target positioning module and a high-confidence updating module are respectively designed. In the tracking process, the target positioning module fuses the gray scale, the direction gradient histogram and the color space characteristics, combines with a characteristic dimension reduction method to train a related filter, and achieves the quick positioning of a target center based on a related filtering algorithm. The high-confidence degree updating module designs a high-confidence degree updating strategy by utilizing the response graph obtained by the target positioning module, namely. the highest response value of the response graph and the average peak correlationenergy (Average Peak-to-Correlation Energy, APCE) are calculated, and scale estimation and model updating are carried out only when the two index values meet conditions at the same time, so that redundant scale estimation operation and filter model updating operation which may introduce noise and cause tracking drift under the condition of low confidence are avoided, and complex scenes such as complicated background and shielding are adapted.

The invention discloses an image matching method based on multiscale Fourier-Mellin transform. The method comprises the following steps of acquiring a reference image I1 and an image to be matched I2 respectively, determining a Gaussian kernel at a pixel position (x, y) in the I2, and calculating a scale space image pixel value L(x, y, t) at the pixel position (x, y) in the I2; further acquiring a scale space image L(t) of the image to be matched I2, and carrying out bilinear interpolation on the L(t) respectively so as to acquire simulation imaging S(t) after the bilinear interpolation; setting a blank image whose size is the same with the size of the L(t), embedding a gray value of each pixel position in the S(t) into the blank image respectively so as to acquire a multiscale image, and calculating a Fourier-Mellin invariant image of the I1 and a Fourier-Mellin invariant image of the multiscale image S'(t) respectively; and calculating a phase correlation matching image between the scale space images L(t) of the I1 and the I2, calculating a convergent-divergent scale estimation value of a pixel position in the I1 and a rotation angle estimation value of the pixel position in the I1 respectively and then acquiring an optimal matching image of the I2.

Provided is an image processing apparatus for fitting a model to image data. The image processing apparatus determines a final parameter set and a final inlier scale of a first model by iteratively performing a model estimation process and an inlier scale estimation process, wherein the model estimation process determines a model parameter set, and the inlier scale estimation process determines an inlier scale by using the determined model parameter set.

Systems and methods permit generation of a digital scan of a user's face such as for obtaining of a patient respiratory mask, or component(s) thereof, based on the digital scan. The method may include: receiving video data comprising a plurality of video frames of the user's face taken from a plurality of angles relative to the user's face, generating a three-dimensional representation of a surface of the user's face based on the plurality of video frames, receiving scale estimation data associated with the received video data, the scale estimation data indicative of a relative size of the user's face, and scaling the digital three-dimensional representation of the user's face based on the scale estimation data. In some aspects, the scale estimation data may be derived from motion information collected by the same device that collects the scan of the user's face.

A manned lunar-landing quality scale estimation method and system are disclosed. The manned lunar-landing quality scale estimation method comprises the four steps of: 1, establishing a quality scale reference database; 2, modularizing an aircraft and a flight process; 3, constructing a flight plan; and 4, solving the quality scale. The manned lunar-landing quality scale estimation system comprises an information reference module, a flight plan construction module and a quality scale solving module; and the three modules are executed in sequence during quality scale estimation, that is to say, firstly the information reference module is constructed, secondly the flight plan construction module is executed and finally the quality scale solving module is executed. According to the manned lunar-landing quality scale estimation method and system, quality data of each subsystem is constructed into the database, and the database can be directly called during quality scale estimation, so that high workload caused by professional and separate calculation of each subsystem is avoided; and the modular thought is further utilized to construct the manned lunar-landing flight plan, so that the construction process is simple, the quality scale can be calculated within a very short time, and the shortcomings of low calculation efficiency, no universality and the like of a conventional method are overcome.