4 results about "Interpolation matrix" patented technology

The present application belongs to the technical field of structure optimization design, and discloses a structure topology optimization design method and device, steps are as follows: (1) dividing a physical design domain into a background grid, based on coordinate information of background nodes of the background grid, a sparse filtering matrix is constructed by using a Gaussian kernel function with row normalization; (2) using the sparse filtering matrix to perform front filtering on design variables on the background nodes to obtain a smooth density field, and then obtaining a level set function field; (3) using an adaptive grid subdivision algorithm to cut and locally adaptively reconstruct the level set function field to obtain an analysis grid; (4) interpolating the level set function field to the analysis grid through a sparse global interpolation matrix from the background grid to the analysis grid to obtain a level set function field; (5) projecting the level set function field to obtain a pseudo-density field on the analysis grid nodes, and then obtaining an optimal topology configuration. The present application realizes good boundary smoothness while improving optimization efficiency.

The present disclosure relates to systems and methods for wide neural networks and sensing array measurement estimation. A full sensing array can be modeled using a wide neural network (BNN). A sensing array can produce a first vector comprising P measurements. The full sensing array can be modeled by producing a second vector comprising Q wave domain measurements using the first vector as follows. The first vector can be multiplied by an interpolation matrix to produce interpolated vectors each comprising Q interpolated values. Applying a transformation function to the interpolated vectors produces a representation vector which can be interleaved to produce Q pooling inputs. The second vector can be produced by performing a plurality of nonlinear pooling operations on the Q pooling inputs, where one of the nonlinear pooling operations produces one of the Q wave domain measurements in the second vector from one of the Q pooling inputs. The interpolation matrix can be produced by training the BNN.

The application discloses a river network hydraulic property interpolation method and related devices, and the application obtains a kernel width parameter vector that is adapted to an interpolation matrix through river network section information (i.e. section mileage and hydraulic properties of the section at different water levels), adopts a Gaussian kernel function, obtains a probability weight matrix, and converts a process of relying on logical judgment and spatial search in traditional interpolation into a weighted fusion process that is completely based on matrix operation, so that conditional judgment branches can be avoided, parallel computing potential of GUP can be fully exerted, the global rationality and stability of the interpolation result can be ensured while the computing efficiency is improved, the unity of the computing performance and the interpolation accuracy is realized, and efficient and high-precision calculation of the river network section property is realized.

This application relates to the field of image processing technology, and further to a method and apparatus for local black level calibration, a computer-readable storage medium, and a computer program product. The method includes: performing temporal accumulation on multiple acquired original images to determine a target image; determining effective pixels of the target image based on the median and standard deviation of multiple target pixels and the mean of each pixel in the target image; determining normalized weights for all effective pixels based on the mask and variance of all effective pixels; determining mesh nodes based on the normalized weights of all effective pixels, the mean of each pixel in the target image, the sampling interpolation matrix, and the regularization coefficient; and performing local black level calibration on the multiple original images based on the mesh nodes. This method significantly eliminates spatial distribution noise from the image sensor and improves image quality.