Compression Perception Reconstruction Algorithm for Interferometric Multispectral Image Based on Machine Learning
Natural signals are generally used in the compressed sensing of multispectral images. However, natural signals do not have sparseness, resulting in the reconstruction time and stability being unable to reach the ideal state. Hence, an interference multispectral image compressed sensing reconstruction algorithm based on machine learning is proposed in this paper. The wavelet transform is used to complete the interference multispectral image coding. The encoded image is used to construct a compressed sensing model. The model is used to find the transformation basis in the natural signal, so that the decomposition coefficients of non-sparse natural signals under this basis are not zero. This solves the problem for non-sparse natural signals. The measurement matrix is designed to ensure the accurate reconstruction of data after effective compression sampling. Then, the SVM decision tree of a machine learning algorithm is used to complete the design of the image reconstruction algorithm. Test results indicate that the proposed algorithm can achieve high stability for both high and low computing platforms. It takes less time and the reconstruction effect is better than that obtained by the existing algorithm. This confirms the usefulness and benefits of the proposed algorithm.
Keywords: multispectral image reconstruction, compressed sensing, machine learning, wavelet transform, transform basis, sparsity
S. Zhang, "Compression Perception Reconstruction Algorithm for Interferometric Multispectral Image Based on Machine Learning", Engineering Intelligent Systems, vol. 30 no. 3, pp. 201-210, 2022.