473 results about "Model fitting" patented technology

Exemplary received signal processing may be based on maintaining a model of received signal impairment correlations, wherein each term of the model is updated periodically or as needed based on measuring impairments for a received signal of interest. An exemplary model comprises an interference impairment term scaled by a first model fitting parameter, and a noise impairment term scaled by a second model fitting parameters. The model terms may be maintained based on current channel estimates and delay information and may be fitted to measured impairment by adapting the model fitting parameters based on the measured impairment. The modeled received signal impairment correlations may be used to compute RAKE combining weights for received signal processing, or to compute Signal-to-Interference (SIR) estimates. Combined or separate models may be used for multiple received signals. As such, the exemplary modeling is extended to soft handoff, multiple antennas, and other diversity situations.

An information processing apparatus according to the present invention includes a three-dimensional model storage unit configured to store data of a three-dimensional model that describes a geometric feature of an object, a two-dimensional image input unit configured to input a two-dimensional image in which the object is imaged, a range image input unit configured to input a range image in which the object is imaged, an image feature detection unit configured to detect an image feature from the two-dimensional image input from the two-dimensional image input unit, an image feature three-dimensional information calculation unit configured to calculate three-dimensional coordinates corresponding to the image feature from the range image input from the range image input unit, and a model fitting unit configured to fit the three-dimensional model into the three-dimensional coordinates of the image feature.

A method of noncontact imaging for performing qualitative and quantitative analysis of wounds includes the step of performing structured illumination of surface and subsurface tissue by both diffuse optical tomography and rapid, wide-field quantitative mapping of tissue optical properties within a single measurement platform. Structured illumination of a skin flap is performed to monitor a burn wound, a diabetic ulcer, a decubitis ulcer, a peripheral vascular disease, a skin graft, and / or tissue response to photomodulation. Quantitative imaging of optical properties is performed of superficial (0-5 mm depth) tissues in vivo. The step of quantitative imaging of optical properties of superficial (0-5 mm depth) tissues in vivo comprises pixel-by-pixel demodulating and diffusion-model fitting or model-based analysis of spatial frequency data to extract the local absorption and reduced scattering optical coefficients.

A system and method is provided for model fitting and registration of objects for 2D-to-3D conversion of images to create stereoscopic images. The system and method of the present disclosure provides for acquiring at least one two-dimensional (2D) image, identifying at least one object of the at least one 2D image, selecting at least one 3D model from a plurality of predetermined 3D models, the selected 3D model relating to the identified at least one object, registering the selected 3D model to the identified at least one object, and creating a complementary image by projecting the selected 3D model onto an image plane different than the image plane of the at least one 2D image. The registering process can be implemented using geometric approaches or photometric approaches.

A system and method for providing lane sensing on a vehicle by detecting roadway lane-markers, where the system employs a surround view camera system providing a top-down view image around the vehicle. The method includes detecting left-side and right-side lane boundary lines in the top-down view image, and then determining whether the lane boundary lines in the image are aligned from one image frame to a next image frame and are aligned from image to image in the top-down view image. If the boundary lines are not aligned, then calibration of one or more of the cameras is performed, and if the lines are aligned, then a model fitting process is used to specifically identify the location of the boundary lines on the roadway.

A computer-implemented simulator models the entire analytic value chain so that data generation, model fitting and strategy optimization are an integral part of the simulation. Data collection efforts, data mining algorithms, predictive modeling technologies and strategy development methodologies define the analytic value chain of a business operation: data→models→strategies→profit. Inputs to the simulator include consumer data and potential actions to be taken regarding a consumer or account. The invention maps what is known about a consumer or an account and the potential actions that the business can take on that consumer or account to potential future financial performance. After iteratively performing simulations using varying inputs, modeling the effect of the innovation on a profit model, the simulator outputs a prediction of the commercial value of an analytic innovation.

A method and system for medical image acquisition are provided, where the method includes acquiring an image of the region of interest, acquiring a model for a region of interest, and fitting the model to the image; the system includes a modeling unit for modeling a region of interest; an acquisition unit in signal communication with the modeling unit for acquiring an image of the region of interest; and a fitting unit in signal communication with the acquisition unit for fitting the model to the image.

In an image classification system (70), a plurality of generative models (30) correspond to a plurality of image classes. Each generative model embodies a merger of a general visual vocabulary and an image class-specific visual vocabulary. A gradient-based class similarity modeler (40) includes (i) a model fitting data extractor (46) that generates model fitting data of an image (72) respective to each generative model and (ii) a dimensionality enhancer (50) that computes a gradient-based vector representation of the model fitting data with respect to each generative model in a vector space defined by the generative model. An image classifier (76) classifies the image respective to the plurality of image classes based on the gradient-based vector representations of class similarity.

The invention relates to a universal road and lane detection system and method. The method comprises: S300, interference elimination and image artifact and irrelevant image part removing are carried out on an input image to obtain a relevant image; S302, lane edge feature or road edge feature is extracted from the relevant image; S304, according to the lane edge feature or road edge feature, fitting is carried out to obtain a synthesized path expression model; S306, the path expression model is tracked to carry out time sequence consistency integration or position consistency integration; and S308, the path expression model corresponds to a global coordinate system. On the basis of the module design, pretreatment is carried out on the image inputted by the camera, the interested relevant image is selected; lane detection or road detection is carried out by combining input information of the laser radar; model fitting is carried out to estimate a road path and the accuracy of path estimation is improved by using the time sequence integration module; and then the image corresponds to the global coordinate system in real time by using a coordinate system correspondence module to improving path estimation accuracy. Therefore, a problem that the existing road estimation model has imperfect functions can be solved.

The invention discloses a model classification system and method, and relates to the technical field of an automobile identification apparatus or method. The classification system comprises a forward snapshot camera, a body scanning camera, a dual-channel video detector and a charging system. The forward snapshot camera and a channel snapshot point are arranged to form an inclination angle of 45 degrees for capturing forward features of a vehicle passing by, and the body scanning camera is arranged at one side of a vehicle channel; and the forward snapshot camera and the body scanning camera are connected with the signal input end of the dual-channel video detector, and video stream vehicle detection analysis software is configured in the dual-channel video detector. The system provided by the invention performs multi-type and multi-dimensional data synchronization, integration and model fitting on information of the same vehicle on the basis of a core layer by use of one high-performance dual-channel video processor through a machine vision algorithm so as to accurately and stably obtain data information needed for accurate vehicle classification in real time and obtain the accurate vehicle classification.

An object is to provide an articulated object position and posture estimation device with reduced calculation cost of model fitting for estimating position and posture and with improved estimation speed. A posture model storage section 2 stores data concerning to a posture model with low-dimensional parameters under movement restraint. The low-dimensional parameters are obtained by performing a principal component analysis on time-series postures of an articulated object frame model corresponding to a predetermined limited movement of an articulated object such as human body. A human body position and posture estimation device 101 generates a image of each posture of the articulated object frame model within postures which can be taken by the posture model in a human body model image generating section 4, performs a matching with an estimation target articulated object image in a position and posture estimation section 10, and thus, estimates a posture.

A frequency domain representation of a whitening filter is made to depend on essentially one unknown, namely, a scaling factor that is based on an estimated ratio of total base station power to the power spectral density (PSD) of inter-cell interference plus noise. In turn, that scaling factor can be computed based on the modeling terms used in a parametric model of the impairment correlations for a received communication signal. Preferably, the model comprises an interference impairment term scaled by a first model fitting parameter, and a noise impairment term scaled by a second model fitting parameter. Further, the scaling factor can be computed by directly estimating total base station transmit power and the PSD of inter-cell interference plus noise. In any case, the whitening filter can be used in whitening a received communication signal in conjunction with channel equalization processing or RAKE receiver processing, for example.

The invention discloses a two-step positioning method of a wireless sensor network based on total least squares, and belongs to the technical field of signal processing. According to the method, a fitting model parameter obtaining module, a distance estimation module and a position positioning module are adopted, the fitting model parameter obtaining module serves the distance estimation module and the position positioning module, the relative distance between a reference node in a practical environment and the terminal and the corresponding RSS (received signal strength) value are collected in advance in real time, and the relative distance and the RSS value are subjected to relationship curve model fitting training to obtain model parameters; the distance estimation module utilizes the real-time collected RSS value for carrying out distance estimation according to the parameters obtained through fitting model training, and different model parameters are selected for estimation according to the RSS value; and the distance obtained by the distance estimation module and the reference node coordinate are used as the input of the position positioning module, and the output of the position positioning module is the estimation value of unknown node coordinates.

An image alignment method includes computationally efficient methods of achieving high-accuracy local motion estimates by using phase correlation. The method also estimates motion reliability that allows a generic robust model fitting algorithm to produce more accurate results while operating much more efficiently. One of three methods are used to determine sub-pel motion estimation with improved accuracy. Each of the sub-pel motion estimation methods uses phase correlation, and are based on fitting computationally efficient 2-D quadratic surfaces to a phase correlation surface. A pre-filter is applied which shapes the phase correlation surface to enable appropriate fitting to the quadratic surface. Bias is also compensated for prior to applying a sub-pel motion estimation method. The method also estimates the reliability of the sub-pel motion estimates determined using phase correlation.

A target in a sequence of measurements is tracked by modeling the target with a switching linear dynamic system (SLDS) having a plurality of dynamic models. Each dynamic model is associated with a switching state such that a model is selected when its associated switching state is true. A set of continuous state estimates is determined for a given measurement, and for each possible switching state. A state transition record is then determined by determining and recording, for a given measurement and for each possible switching state, an optimal previous switching state, based on the measurement sequence, where the optimal previous switching state optimizes a transition probability based on the set of continuous state estimates. A measurement model of the target is fitted to the measurement sequence. The measurement model is the description of the influence of the state on the measurement. It couples what is observed to the estimated target. Finally, a trajectory of the target is estimated from the measurement model fitting, the state transition record and parameters of the SLDS, where the estimated trajectory is a sequence of continuous state estimates of the target which correspond to the measurement sequence. The set of continuous state estimates is preferably obtained through Viterbi prediction. The optimal previous switching state can be an optimal prior switching state, or can be an optimal posterior switching state.

Described is a solution for an unlabeled target domain dataset challenge using a domain adaptation technique to train a neural network using an iterative 3D model fitting algorithm to generate refined target domain labels. The neural network supports the convergence of the 3D model fitting algorithm and the 3D model fitting algorithm provides refined labels that are used for training of the neural network. During real-time inference, only the trained neural network is required. A convolutional neural network (CNN) is trained using labeled synthetic frames (source domain) with unlabeled real depth frames (target domain). The CNN initializes an offline iterative 3D model fitting algorithm capable of accurately labeling the hand pose in real depth frames. The labeled real depth frames are used to continue training the CNN thereby improving accuracy beyond that achievable by using only unlabeled real depth frames for domain adaptation.

The invention discloses a method for positioning characteristic points in a facial image and belongs to the technical field of signal processing. The method for positioning the characteristic points in the facial image comprises a model training stage and an image searching stage. At the model training stage, a sufficient number of normal facial images and facial images under different light conditions are selected to serve as a training set, the positions of characteristic points of each training image are accurately marked manually, and a moving shape model and a local texture model are established. The model training stage just needs to be executed once. At the image searching stage, the accurate positions of pupils of a target image are firstly determined to initialize the initial position of the moving shape model, and then the model is fitted to the accurate target characteristic point positions through iteration.

Signal impairment correlations for multiple signals in a received multipath signal are constructed by fitting parametric models associated with each high-data-rate signal in the multipath against measured impairment correlations. The estimated model fitting parameters are applied to form impairment correlation estimates for all signals. The models comprise a separate impairment covariance matrix scaled by a model fitting parameter for each high-data-rate signal and a noise covariance matrix scaled by a noise element model fitting parameter. The model fitting parameters may be estimated by a least-squares formulation and applied to form impairment correlation estimates for all signals of interest. The resulting impairment correlation estimates may be provided to G-RAKE receivers or joint scaling demodulators to demodulate the signals while suppressing interference from the high-data-rate signals.

The invention discloses a virtual fitting mirror system. The system comprises a skeleton identification system which identifies and is bound to a body skeleton firstly entering an identification scope and controls a stereo model of a virtual scene to act through body motion; a virtual fitting system which is used for fitting of a somatosensory model, i.e., selecting clothes through palm action and carrying out the fitting through the somatosensory model; an assistant matching system which is used for fitting of a virtual model, i.e., establishing the virtual model, selecting the clothes through a touch point-selection mode and performing the fitting through the virtual model; a clothes self-selection system which is used for trying self-selected clothes, i.e., inputting bar codes to schedule the corresponding clothes so as to perform fitting on the somatosensory model; and an intelligent push system which is used for pushing entity clothes, i.e., pushing the entity clothes through selecting the clothes via an intelligent push wardrobe. According to the invention, online and offline somatosensory model fitting, virtual model fitting and human body fitting can be provided, and the problem of insufficient seller quantities and fitting room quantities of offline entity shops can be solved to a quite large degree.

The invention belongs to the technical field of insole making and discloses an individualized 3D printed insole and a making method thereof. The method comprises the following steps: (1) obtaining basis data of foot size to establish a foot three-dimensional model; partitioning the sole, analyzing the health condition of the foot according to relevant data of each zone of the sole, and determining the three-dimensional curve of each zone and the three-dimensional curve of the foot; (2) dividing the insole into a bottom layer, a middle layer and an upper layer, wherein the bottom layer is made from thermoplastic polymer hard material, the middle layer is made from modified polyurethane elastomer, and the top layer is made from nylon plant fiber composite material; (3) deriving an insole structure model fitting the foot according to the foot three-dimensional model, relevant data of each zone of the sole, the health condition of the foot, and individual requirements; (4) carrying out 3D printing on the insole. The method disclosed by the invention is economical, convenient and efficient, individualized adjustment on the performance of local materials can be carried out for different zones of the sole, and insoles capable of meeting the requirements of different people can be produced.

The invention discloses a method for identifying an electroencephalogram image based on a deep convolutional neural network. The method comprises the following steps: carrying out baseline eliminationpreprocessing on a collected motor imagery electroencephalogram signal; Dividing each lead signal into a plurality of time windows, and carrying out fast Fourier Transform on each window MI-EEG model, carrying out fast Fourier inverse transform on the EEG models respectively, and calculating corresponding time domain power values of the EEG models; Calculating a mean value of the time domain power values obtained by each window to obtain time domain power characteristics; Performing interpolation imaging on the extracted three-frequency-band power characteristics in a data matrix to obtain apseudo RGB image of the MI-EEG signal; Designing the DCNN model into five segments of convolution, and after each segment of convolution is finished, replacing a maximum pooling layer with a convolution layer to carry out data dimension reduction; And performing evaluation on the test set by using the trained DCNN model to complete a classification test. MI-EEEG images have the advantages in the aspect of feature expression, and are matched with 30 layers of DCNN with higher model fitting capability, which has great significance for the improvement of the MI- EEG signal feature expression andclassification precision .

An image alignment method includes computationally efficient methods of achieving high-accuracy local motion estimates by using phase correlation. The method also estimates motion reliability that allows a generic robust model fitting algorithm to produce more accurate results while operating much more efficiently. One of three methods are used to determine sub-pel motion estimation with improved accuracy. Each of the sub-pel motion estimation methods uses phase correlation, and are based on fitting computationally efficient 2-D quadratic surfaces to a phase correlation surface. A pre-filter is applied which shapes the phase correlation surface to enable appropriate fitting to the quadratic surface. Bias is also compensated for prior to applying a sub-pel motion estimation method. The method also estimates the reliability of the sub-pel motion estimates determined using phase correlation.

The invention relates to the technical field of application of optoelectronic products and particularly relates to a method for detecting a dim small moving target under a downward-looking complicated background. In order to solve the problem of detection and reorganization of the infrared dim small moving target under a complicated ground feature background under a long-distance ground feature background, the method takes the problem of real-time performance into full account, firstly images are aligned, a method of comprehensively estimating time domain and space domain is used for the complicated background with ground features to accurately estimate a static background, difference is performed between the static background and the current image to obtain an image with the background seriously inhibited, then an edge inhibition method is used to eliminate the strong edge in the difference image, the target is divided by using a small facial model fitting process according to the Gaussian-like distribution of a small target gray level, and finally target marking and feather extracting are conducted, the interference of random noise is eliminated through multiframe relevance to finally extract the correct target. The method can accurately detect the dim small moving target.

A system and method is provided for model fitting and registration of objects for 2D-to-3D conversion of images to create stereoscopic images. The system and method of the present disclosure provides for acquiring at least one two-dimensional (2D) image (202), identifying at least one object of the at least one 2D image (204), selecting at least one 3D model from a plurality of predetermined 3D models (206), the selected 3D model relating to the identified at least one object, registering the selected 3D model to the identified at least one object (208), and creating a complementary image by projecting the selected 3D model onto an image plane different than the image plane of the at least one 2D image (210). The registering process can be implemented using geometric approaches or photometric approaches.

The invention relates to a foot-shaped three-dimensional surface reconstruction method based on image segmentation and grid subdivision, comprising the steps of: carrying out statistics and analysis on a shoe last sample set to obtain a statistic deformation model, using a plurality of cameras to obtain images of a foot, then using the statistic model to fit the foot-shaped images, obtaining a sparse grid model, then segmenting image characteristic points from all the images, reducing the planar characteristic points into spatial points, finally using the spatial characteristic points subdivide the grid model and carrying out the subdivision iteratively, thus obtaining a foot-shaped model consistent to a target object. In the foot-shaped three-dimensional surface reconstruction method, marking points do not need to be arranged on the foot, high-accuracy laser measurement equipment is also not needed, artificial participation is not needed, only the cameras and a computer and other devices are used, and the reconstruction risk is finished by software in a full-automatic manner. A reconstruction network can be consistent to a target foot shape automatically, can capture the detail characteristics in the images automatically and can adapt to information provided by the images automatically.

The invention discloses a method and device for measuring microstructure morphology based on spectral modulation depth coding. An element to be measured and a spatial light modulator are conjugated atthe center wavelength of the spectral range used in the measurement; a beam folding coupler, the spatial light modulator, a collimating and expanding lens, a beam splitter, an axial non-achromatic microscope objective lens, an imaging lens and a color camera are in a common optical path structure. During the measurement, the pre-calibration of the corresponding relation of the 'spectral-depth' isfirstly carried out on a system device, and each frame monochromatic shift-phase fringe pattern reflected by the element to be measured is acquired by a measuring device, so as to obtain the modulation depth distribution of each monochromatic light fringe pattern related to the surface shape of the element to be measured, and the coded image is obtained; the 'spectrum-modulation depth' relation curve of each point on the surface to be measured is determined by using Gaussian, quasi-Gaussian or spline model fitting, the depth information of each point on the surface to be measured is demodulated, and the fast and precise measurement of the microstructure topography of the three-dimensional topography distribution of the element to tested, without mechanical scanning or contact, is completed.

The invention discloses an alignment signal processing method based on amplitude and phase probed separately. The method breaks a rough alignment signal from an alignment signal into sections and summarizes the same as well estimates peak value position; determines a rough alignment position of marker by rough alignment signal model fitting; simplifies and summarizes the accurate alignment signal from the alignment signal and determines peak value point of the accurate alignment signal by accurate alignment signal model fitting; and synthesizes the rough alignment position and the accurate alignment signal peak value point to determine a final alignment position. The alignment signal processing method provided by the invention can effectively improve alignment signal processing efficiency to acquire correct alignment position.

The invention discloses a high-texture image classification method and a system based on visual saliency detection, wherein, the method comprises the following steps of: segmenting a saliency region through an image pixel saliency value detection method of color contrast; the texture features and gradient features of salient regions are extracted by the complete local binary mode operator and thedirectional gradient histogram algorithm, and the image details are represented by the effective tandem fusion strategy. And classifying the extracted fusion vector by the nearest neighbor classifierto obtain the recognition rate. This method utilizes the visual attention model fitting human visual mechanism to locate the salient target region in the image, in order to get rid of the redundant background information, the high-texture image is segmented, and then the feature of the salient region is extracted and the image classification is completed, so that the different textures in the image (field of view) are automatically separated, which makes the subsequent scene description or target recognition possible.

A computer-implemented method of analysis of melt curve data comprising the steps of: (a) parameter model fitting the melt curve data; (b) performing a principal component analysis of the melt curve data; and (c) utilizing the principal components for clustering the melt curve data into groups. Such a method allows an efficient analysis of variations in melt curve shape and position and allows to statistically quantify these variations for both supervised and unsupervised data sets. Current melt analysis methods neither allow for statistical measures of each unknown nor do they allow for the determination of unsupervised data sets (i.e., unknown number of groups present). Particularly, the method according to the invention can be advantageous for identifying a specific sequence of dsDNA in a sample after performing a polymerase chain reaction (PCR).

The invention provides a vehicle three-dimensional detection method based on a model fitting algorithm. The method mainly involves vehicle dimension estimation, vehicle model fitting and two stage refining of convolutional neural network. The process of the method is that an original image is transmitted to a two-dimensional detection network, the network generates a two-dimensional bounding box for candidate vehicles in an image plane; a set of three-dimensional points that fall into the two-dimensional bounding box after projection are selected; the model fitting algorithm is performed on detection of a three-dimensional position and the three-dimensional bounding box of a vehicle by utilizing the set; the points suitable for the three-dimensional bounding box are input; the two stage refining convolutional neural network is designed; the detected three-dimensional bounding box is aligned with point cloud; and final three-dimensional box regression and classification are performed. The model fitting algorithm proposed by the method can provide three-dimensional information by utilizing the advantages of either two-dimensional detection network, a more efficient model fitting process is achieved, and the capability and detection precision of three-dimensional vehicle detection are improved.