54results about How to "Guaranteed reconstruction accuracy" patented technology

The invention discloses a three-dimensional human body rapid reconstruction method based on a simple measuring suit. For each sample human body in the existing human body database, extract characteristic parameters and characteristic curves as human body characteristics to form a human body characteristic database; use PCA method to process and construct human body shape space; establish a mapping model between human body characteristics and human body shape space through neural network ;Measure the characteristic parameters of the human body to be measured, and use the simple measuring suit to obtain the characteristic curve of the human body to be measured; input the characteristic parameters and characteristic curves of the human body to be measured into the mapping model to obtain the corresponding feature vectors of the human body to be measured in the human body shape space eigenvalues, and then restore and reconstruct to obtain a 3D human body model. In the present invention, the characteristic curve of high spatial dimension provides abundant geometric information of the human body surface, and the multi-layer neural network accurately reflects the mapping model between the human body features and the human body shape space, thereby improving the accuracy and reliability of the three-dimensional reconstruction of the human body.

The invention discloses a regional depth edge detection and binocular stereo matching-based three-dimensional reconstruction method, which is implemented by the following steps: (1) shooting a calibration plate image with a mark point at two proper angles by using two black and white cameras; (2) keeping the shooting angles constant and shooting two images of a shooting target object at the same time by using the same camera; (3) performing the epipolar line rectification of the two images of the target objects according to the nominal data of the camera; (4) searching the neighbor regions ofeach pixel of the two rectified images for a closed region depth edge and building a supporting window; (5) in the built window, computing a normalized cross-correlation coefficient of supported pixels and acquiring the matching price of a central pixel; (6) acquiring a parallax by using a confidence transmission optimization method having an acceleration updating system; (7) estimating an accurate parallax by a subpixel; and (8) computing the three-dimensional coordinates of an actual object point according to the matching relationship between the nominal data of the camera and the pixel andconsequently reconstructing the three-dimensional point cloud of the object and reducing the three-dimensional information of a target.

The invention discloses a method for rebuilding front and back oval edges of an aerial blade based on error control, which is used for solving the technical problem of poor anti-noise performance existing in the conventional high-precision rebuilding method for a front edge of a blade. According to the technical scheme, the method comprises the following steps of: selecting region points covering the front edge or the back edge of the blade; sequencing the region points; performing spline fitting on the sequenced region points for three times; performing equal-arc-length encryption discretion on fitted spline curves, and marking beginning and end discrete points as P0i and P1i respectively; fitting discrete points at a P0iP1i section by using a least square method, and marking beginning and end discrete points on the intersection of the P0iP1i section and a fitting oval as P0(i+1) and P1(i+1) respectively; calculating an average error epsiloni ranging from the discrete points to an ith fitting oval; and setting an iteration control error as epsilon till the epsiloni is less than or equal to epsilon to finish iteration. According to the method, the front and back oval edges of the blade are rebuilt by adopting an error control-based iterative fitting method on the basis of fitting through the least square method, so that points on a non-oval circular arc in the selected region are prevented from participating in fitting, and the anti-noise performance is improved.

The present application belongs to the technical field of image processing, and provides an image reconstruction method, apparatus, device and storage medium. The method obtains the sampling data of the target object; the sampling data is input into the deep learning network after training for processing, and the reconstructed image corresponding to the sampling data is obtained, wherein the sampling data is obtained based on a preset sampling mode in a non-Cartesian coordinate system. Under-sampling frequency-domain data; compared to the prior art technical solution for image reconstruction based on non-uniform fast Fourier transform on under-sampling frequency-domain data in a non-Cartesian coordinate system, the deep learning network in the embodiment of the present application undergoes image reconstruction. Pre-training can directly reconstruct the corresponding image according to the input undersampling frequency domain data in the non-Cartesian coordinate system, without manual selection / adjustment of parameters such as scale factors, which improves the undersampling frequency in the non-Cartesian coordinate system. Reconstruction speed of domain data.

The invention discloses a reconstruction method and device for target electromagnetic-scattering-characteristic data, and relates to the technical field of radar imaging. The method includes the stepsthat a three-dimensional echo signal model of a target is built, and the three-dimensional echo signal model is converted into a state-space equation; according to scattering echo data of the target,a generalized Hankel matrix is built, and the state-space equation is solved based on the Hankel matrix to determine a parameter estimating value of the target scattering center; according to the parameter estimating value of the target scattering center, electromagnetic-scattering-characteristic data of the target is reconstructed. Through the steps, the reconstruction accuracy of the target electromagnetic-scattering-characteristic data is guaranteed, and efficient compression and real-time loading of the full-space target electromagnetic-scattering-characteristic data can also be achieved.

The invention discloses a vehicle-mounted rail frequency shift signal rapid compression method based on compressed sensing. The vehicle-mounted rail frequency shift signal rapid compression method is used for rapid compression of rail frequency shift signals in LKJs, DMSs and vehicle-mounted subjectivization cab signal systems on ordinary-speed trains and high-speed trains. The method includes the steps of firstly, calculating the sparsity of a vehicle-mounted rail frequency shift signal to be compressed, and determining whether to conduct sparsity conversion on the vehicle-mounted rail frequency shift signal to be compressed or not according to the sparsity; secondly, compressing a sparsity coefficient through a gauss random matrix to obtain compressed data; thirdly, effectively recovering the rail frequency shift signal just through a compressed sensing reconstruction method when uncompressing the compressed data. According to the method, the compression ratio of the rail frequency shift signal is high, the instantaneity is high, the storage loads of the vehicle-mounted LKJs, the vehicle-mounted DMSs and the vehicle-mounted cab signal systems can be effectively reduced, the bandwidth requirement of vehicle-ground wireless transmission of the vehicle-mounted rail frequency shift signal is lowered, the transmission time delay of vehicle-ground wireless transmission of the vehicle-mounted rail frequency shift signal is reduced, and the method has wide application prospects in high-speed rail vehicle-mounted signal real-time monitoring.