Non-local low-rank conversion domain and full-connection tensor decomposition image reconstruction method and device

Abstract

The invention discloses a non-local low-rank conversion domain and full-connection tensor decomposition image reconstruction method and device. The method comprises the following steps: S1, inputting a to-be-restored image; s2, constructing a tensor decomposition model: S2.1, segmenting the input image to obtain non-local tensor blocks; s2.2, introducing the non-local tensor block into a B spline conversion domain to obtain a conversion domain form of the non-local tensor block; s2.3, through the non-local tensor blocks, constructing a non-local similar tensor block group; s2.4, combining full-connection tensor decomposition, and constructing a full-connection tensor decomposition factor; s2.5, constructing a low-rank tensor completion model, and performing optimization according to the steps S2.1 to 2.4 to obtain a non-local low-rank conversion domain and full-connection tensor decomposition-based model; and S3, constructing an image restoration model, obtaining a to-be-restored image, and obtaining a restored image through a restored image tensor block group obtained through the tensor decomposition model. In the spectral image restoration process, the whole image reconstruction is more accurate.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an image reconstruction method and device based on non-local low-rank transformation domain and fully connected tensor decomposition. Background technique [0002] Due to factors such as manufacturing processes, device aging, or transmission errors, pixels are lost during the capture and generation of high-dimensional image data. Low Rank Tensor Completion (LRTC) restores lost elements based on the low rank of the data set. Matrix completion is a second-order tensor completion method, usually assuming that the matrix is ​​low-rank, and using this as a constraint to minimize the difference between the given incomplete matrix and the estimated matrix. However, when the data to be analyzed has a complex structure, using a matrix to describe high-dimensional data has problems such as curse of dimensionality, overfitting, and incomplete data structure information. Therefore...

AI Technical Summary

Problems solved by technology

However, methods based on deep learning require a large number of labeled samples, which are difficult to obtain and require high computing power, so the research and application of small-sample traditional methods are still necessary, and there is a lot of room for improvement

Images