8 results about "Hyperspectral image compression" patented technology

The invention discloses a hyperspectral image compression method based on an SDMNet network model, and relates to the field of hyperspectral image compression, in particular to a hyperspectral image compression method. The invention aims to solve the problems that a large number of frequency domain features are lost and a reconstructed image is blurred in an existing image compression method. The method comprises the following steps: acquiring an original hyperspectral image block; an SDMNet network model is constructed; the SDMNet network model represents a network model based on spectrum difference reserved header and Mamba pyramid feature enhancement; the SDMNet network model sequentially comprises a spectrum difference reservation header SDPH, a C block 1, a C block 2, a Mama pyramid feature enhancement module MPFM, a C block 3, a coding and decoding module, an R block 3, a Mama pyramid feature enhancement module MPFM, an R block 2, an R block 1 and a spectrum difference reservation header SDPH; and compressing the original hyperspectral image block to be measured based on the trained SDMNet network model, and outputting compressed data.

The invention discloses a hyperspectral image compression network and compression method based on multi-scale spectrum and spatial feature enhancement, and mainly solves the problems of insufficient hyperspectral image spectrum modeling, insufficient spatial feature extraction and high network complexity in the prior art. The network comprises a main encoder, a main decoder, a super-prior encoder, a super-prior decoder and an entropy model. The main encoder comprises a spectral attention gating data unit, a convolution unit and a multi-scale spatial adaptive feature attention enhancement unit, and is used for converting an input image into potential representation and removing spatial and spectral redundancy; the main decoder and the main encoder are symmetrical in structure; the super-prior encoder comprises a convolution layer and an activation layer and is used for extracting auxiliary information from the output of the main encoder; the super-prior decoder and the super-prior encoder are symmetrical in structure; the entropy model is Gaussian distribution based on output parameters of a super-prior decoder. After the network is trained, lossy compression of a hyperspectral image can be realized. The method reduces the network complexity and spectral distortion, improves the reconstruction quality of complex ground feature details, and is suitable for earth observation, meteorological monitoring and the like.

The application discloses a hyperspectral image compression reconstruction method based on Hankel rank increasing tensor decomposition, and belongs to the technical fields of image processing and spectral remote sensing, and comprises the following steps: step 1, a basic hyperspectral image compression reconstruction model based on a regular term constraint is constructed to obtain a compression measurement value of the hyperspectral image; step 2, a high-order tensor based on a Hankel transformation is constructed; step 3, a regular term based on Hankel rank increasing tensor decomposition is constructed for the high-order tensor obtained in step 2; step 4, the regular term of the Hankel rank increasing tensor decomposition constructed in step 3 is substituted into a sparse regular term of the basic compression reconstruction model in step 1 to form a final hyperspectral image compression reconstruction model; and step 5, an alternating optimization method is used to solve the hyperspectral image compression reconstruction model constructed in step 4. The application can effectively solve the problems of large data volume of the hyperspectral image, storage and transmission difficulties and the problem that the existing compression reconstruction method ignores multi-dimensional structure information.

The application is based on a hyperspectral image compression and reconstruction method based on three-dimensional collaborative attention, comprising selecting a hyperspectral image dataset for preprocessing, dividing the preprocessed dataset into a training set, a validation set and a test set, establishing a hyperspectral image compression network model composed of an encoder and a decoder, the encoder comprising a three-dimensional collaborative attention module, the decoder being based on a feature enhancement backbone architecture and being configured with a feature refining module, training, validating and testing the hyperspectral image compression network model to obtain an optimized hyperspectral image compression network model, inputting an image to be compressed into the optimized hyperspectral image compression network model to realize compression and reconstruction of the hyperspectral image. The application realizes fusion expression of spatial-spectral-channel information in the encoding stage by introducing a three-dimensional collaborative attention mechanism, thereby improving the semantic integrity and discriminability of the compressed features as a whole and improving the compression performance of the model.

The application provides a hyperspectral image compression reconstruction method and system guided by a spectral gradient, the method comprising collecting a hyperspectral image and performing preprocessing on the hyperspectral image to construct a hyperspectral image dataset; constructing an end-to-end spectral gradient guided network, the spectral gradient guided network comprising an encoder and a decoder; the spectral gradient Transformer module is arranged in the encoder and the decoder; the hyperspectral image dataset is used to perform end-to-end training on the spectral gradient guided network; the hyperspectral image to be compressed is input into the trained spectral gradient guided network, the encoder is used to encode the hyperspectral image to be compressed to obtain compressed encoding data; the decoder is used to decode the compressed encoding data to obtain a reconstructed hyperspectral image. The application can realize efficient compression and high-fidelity reconstruction of urban scene hyperspectral images, and provides reliable technical support for engineering landing of hyperspectral imaging technology in the field of urban intelligent driving.

The invention discloses a hyperspectral image compression network and compression method based on a 3D convolution set and a causal entropy model, and mainly solves the problems that the spatial-spectral features of a hyperspectral image cannot be jointly extracted and balanced and an entropy model depends on a fixed causal relationship in the prior art. The method comprises an encoder based on a 3D convolution set, an asymmetric decoder based on the 3D convolution set, a super-prior encoder, a decoder and a context causal adaptive entropy model. The encoder comprises a 3D convolution set attention residual block, a spectrum self-attention module and a convolution module, and is used for converting input into potential features and balancing spatial spectrum features; the asymmetric decoder comprises a 3D convolution set attention residual block, an up-sampling 3D convolution set module, a spectrum self-attention module and a convolution module, and is used for recovering potential features into images; the super-prior encoder comprises a convolution layer and an activation layer and is used for extracting statistical information output by the encoder; after the network is trained, lossy compression of the hyperspectral image can be realized. The method can improve the image space and spectrum reconstruction quality, reduces the bit number required by coding, and can be used for environmental monitoring and resource exploration.

The invention discloses a hyperspectral image compression and reconstruction method based on graph tensor structure sparse constraint, and belongs to the technical field of image processing and spectral remote sensing, and the method comprises the steps: 1, constructing a basic hyperspectral image compression and reconstruction model based on regular term constraint, and obtaining a compression measurement value of a hyperspectral image; 2, constructing a sparse constraint regular term based on a tensor structure; step 3, constructing a non-linear smooth regular term based on the graph tensor; 4, fusing a tensor structure sparse constraint regular term and an image tensor nonlinear smooth regular term, replacing a sparse regular term of the basic hyperspectral image compression and reconstruction model in the step 1, and forming a final hyperspectral image compression and reconstruction model; and 5, solving the hyperspectral image compression reconstruction model constructed in the step 4 by adopting an alternating optimization method, and finally outputting a reconstructed hyperspectral image. The method can effectively solve the problems that a hyperspectral image is large in data size and difficult to store and transmit, and an existing compression reconstruction method neglects multi-dimensional structure information.

The invention discloses a hyperspectral image compression and reconstruction method based on Hankel rank increase tensor decomposition, which belongs to the technical field of image processing and spectral remote sensing, and comprises the following steps: step 1, constructing a basic hyperspectral image compression and reconstruction model based on regular term constraint, and obtaining a compression measurement value of a hyperspectral image; step 2, constructing a high-order tensor based on Hankel transformation; step 3, aiming at the high-order tensor obtained in the step 2, constructing a regular term based on Hankel rank increase tensor decomposition; 4, replacing the sparse regular term of the basic compression reconstruction model in the step 1 with the regular term of Hankel rank increase tensor decomposition constructed in the step 3 to form a final hyperspectral image compression reconstruction model; and 5, solving the hyperspectral image compression reconstruction model constructed in the step 4 by adopting an alternating optimization method. The method can effectively solve the problems that a hyperspectral image is large in data size and difficult to store and transmit, and an existing compression reconstruction method neglects multi-dimensional structure information.