Rong Qu

TL;DR: In this article, a chaos depth wavelet network is constructed and used for extracting features of the data automatically, richer feature expressions are obtained, manual participation is avoided, uncertain factors are eliminated, and the classification accuracy is improved.

TL;DR: In this paper, a high-resolution SAR image change detection method based on curve wave SAE was proposed, where a training data set is constructed according to two registered two-time phase SAR images of one same region, and normalization is further carried out.

TL;DR: In this article, a dual-channel deep convolutional generative adversarial network (DCGAN) and feature fusion is proposed for multispectral image classification, which combines the feature fusion, extracts avariety of multi-spectral high-level feature information in multiple directions, enhances the feature characterization ability, and improves the classification effect.

TL;DR: The modelling framework presented in this paper can be extended to much wider spectra of scheduling problems concerning patients with different long-term conditions in future work.

TL;DR: In this paper, a polarimetric SAR image classification method based on a weighted dense network is proposed, which comprises the steps of: (1) building a weighted density network; (2) selecting to-be-classified polarIMetric SAR images; (3) performing filtering; (4) obtaining scattering features; (5) forming a three-dimensional feature matrix from the scattering feature values of the to be-classified SAR image; (6) generating training data sets and testing data sets; (7) classifying the training data set by using the weighted densen