There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality.

Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

I and i

다중혈관 관상동맥 환자에서 y-문합을 이용하여 양쪽 내흉동맥만을 사용한 우회술의 조기 성적

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

An overview of the self-organizing map algorithm, on which the papers in this issue are based, is presented in this article.

The Self-Organizing Map

The realtime implementation is discussed for several detection and classification techniques: Orthogonal Subspace Projection (OSP), Filter Vector Algorithm (FVA), Generalized Likelihood Ratio Test (GLRT), RX algorithm, Constrained Energy Minimization (CEM), Target Constrained Interference Minimization Filter (TCIMF), and Constrained Linear Discriminant Analysis (CLDA). Two data dimensionality limitations are met in realtime processing. One is the number of classes to be classified cannot be larger than data dimensionality, i.e., the number of spectral bands (for some techniques), and the other is the number of independent pixel vectors used for processing must be larger than the number of bands for a data sample correlation or covariance matrix with full rank (for some techniques). In this paper, we present methods to take care of these two limitations: the former is solved by generating artificial band images to expand the data dimensionality, while the latter is solved by using a positive definite correlation matrix as initial matrix. Experiments using hyperspectral data and multispectral data demonstrate the effectiveness of these methods.

Real-time online processing for remote sensing imagery

Hyperspectral remote sensing images have high spectral resolution that enables accurate object detection, classification, and identification. But its vast data volume brings about problems in data transmission, data storage, and data analysis. How to reduce the data volume while keeping the important information for the following data exploitation is a challenging task. Principal Components Analysis (PCA) is a typical method for data compression, which re-arranges image information into the first several principal component images in terms of variance maximization. But variance is not a good criterion to rank images. Instead, signal-to-noise ratio (SNR) is a more reasonable criterion, and the resulting PCA is called Noise Adjusted Principal Components analysis (NAPCA). It is also known that interference is a very serious problem in hyperspectral remote sensing images, induced by many unknown and unwanted signal sources extracted by hyperspectral sensors. Signal-to-interference-plus-noise (SINR) was proposed as a more appropriate ranking criterion. The resulting PCA is referred to as Interference and Noise Adjusted PCA (INAPCA). In this paper, we will investigate the application of INAPCA to hyperspectral image compression, and compare it with the PCA and NAPCA-based compression. The focus is the analysis of their impacts on the following data exploitation (such as detection and classification). It is expected that using NAPCA and INAPCA higher detection and classification rates can be achieved with a comparable or even higher compression ratio. The results will be compared with popular wavelet-based compression methods, such as JPEG 2000, SPIHT, and SPECK.

Interference and noise adjusted principal components analysis for hyperspectral remote sensing image compression

This paper investigates an efficient framework for the fusion of hyperspectral and LiDAR-derived digital surface model to improve classification performance, where collaborative representation based classifier is chosen due to its high computational efficiency with an analytical solution. Local binary pattern (LBP) and extinction profile (EP) features are extracted from both the sources, which include different spatial attributes. Then the derived spatial features are fed to a collaborative representation-based classifier with Tikhonov regularization (CRT) to produce representation residuals. Weighted residuals are calculated, and class label is assigned according to the minimal residual class to generate the classification map ultimately. To improve classification accuracy, the kernel CRT (KCRT) is used and residual fusion (RF) is conducted for the representation residuals from different sources and features. In this paper, spatial filtering for KCRT-RF is investigated. Experimental results demonstrate that a guided filter can help improve the fusion performance of KCRT-RF without significantly increasing computing cost.

Hyperspectral and LiDAR data fusion using collaborative representation

Traditional clustering-based band selection (BS) methods treat each band as individuals, and selection is conducted by enlarging the difference between clusters, which leads to the loss of band interaction and information saliency evaluation. In this article, we propose a BS method named rank-aware generative adversarial network (R-GAN) to address these problems. First, centralized reference feature extraction (FE) with GAN aids R-GAN to combine interpretability and interband relevance. Then, the reference feature is refined with the saliency estimation provided by the rank-aware strategy. According to data characteristics, there are two versions of rank computation including tensor and matrix. Finally, the structural similarity index measurement (SSIM) maps the saliency to the original data space to obtain the final BS result. Extensive comparison experiments with popular existing BS approaches on five hyperspectral images (HSIs) datasets show that the proposed R-GAN can address spectral saliency effectively and select more informative band subsets, which outperforms other competitors for both detection and classification tasks. For example, on the SD-1 dataset, the ten bands selected by R-GAN achieve 0.982 ± 0.003 with an improvement of 13.7% in the area under the curve (AUC) value of anomaly detection performance. The peaked accuracy surpasses the baseline by 0.46% for the classification on the PaviaU dataset.

Rank-Aware Generative Adversarial Network for Hyperspectral Band Selection

Hyperspectral image (HSI) and light detection and ranging (LiDAR) data fusion have been widely employed in HSI classification to promote interpreting performance. In the existing deep learning methods based on spatial–spectral features, the features extracted from different layers are treated fairly in the learning process. In reality, features extracted from the continuous layers contribute differentially to the final classification, such as large tracts of woodland and agriculture typically count on shallow contour features, whereas deep semantic spectral features have meaningful constraints for small entities like vehicles. Furthermore, the majority of existing classification algorithms employ a patch input scheme, which has a high probability to introduce pixels of different categories at the boundary. To acquire more accurate classification results, we propose a spatial–spectral saliency reinforcement network (Sal2RN) in this article. In spatial dimension, a novel cross-layer interaction module (CIM) is presented to adaptively alter the significance of features between various layers and integrate these diversified features. Moreover, a customized center spectrum correction module (CSCM) integrates neighborhood information and adaptively modifies the center spectrum to reduce intraclass variance and further improve the classification accuracy of the network. Finally, a statistically based feature weighted combination module is constructed to effectively fuse spatial, spectral, and LiDAR features. Compared with traditional and advanced classification methods, the Sal2RN achieves the state-of-the-art classification performance on three open benchmark datasets.

Qian Du

Papers

Real-time online processing for remote sensing imagery

Interference and noise adjusted principal components analysis for hyperspectral remote sensing image compression

Hyperspectral and LiDAR data fusion using collaborative representation

Rank-Aware Generative Adversarial Network for Hyperspectral Band Selection

Sal²RN: A Spatial–Spectral Salient Reinforcement Network for Hyperspectral and LiDAR Data Fusion Classification