Most of the existing anomaly detection (AD) approaches for hyperspectral images (HSIs) usually achieve high detection accuracy at the cost of high computational complexity. To meet the needs of actual detection scenarios (efficiency, robustness, and accuracy), this letter introduces a fast and robust AD algorithm via subfeatures grouping and binary accumulation (SFBA) for HSIs. We propose a spatial–spectral anomaly scoring strategy to improve detection accuracy. A number of spectral subfeatures of HSI are selected and divided into $N$ groups as the detection input information and the proposed scoring strategy is carried out for $N$ group data, respectively, to improve the detection efficiency. Then binary accumulation is introduced to improve the robustness of AD algorithm by accumulating detection results of each group. Our proposed detection algorithm was compared with the existing algorithms on real hyperspectral datasets, thereby verifying its strong robustness and low computational complexity simultaneously.

Rapid detection of the new class of hypersonic targets (HTs) presenting unknown military threats in space-based surveillance will guarantee aerospace security. This paper proposes an unsupervised subclass definition and an efficient isolation forest based on an anomalous hyperspectral feature selection (USD-EiForest) algorithm to detect the new class of never-before-seen HTs under emerging hyperspectral sample streams. First, we reveal that the hyperspectral features (HFs) of the new class of HTs have no anomaly characteristics when compared to the globally observed samples while having prominent anomaly characteristics when compared to the subclasses of observed samples. Second, an unsupervised subclass definition method adapted to HTs is utilized to classify the observed samples into several subclasses. Then, an efficient isolation forest is designed to determine whether the data stream sample in each subclass indicates anomaly features that mark the detection of the new class of hypersonic targets (DNHT). Finally, we experiment on the simulated hyperspectral HTs data sets considering the RAM-C II HT as the observed samples and the HTV-2 HT as the unknown samples. The results suggest that the performance of our proposal has competitive advantages in terms of accuracy and detection efficiency.

An aircraft that moves through the atmosphere at hypersonic speed is covered by plasma sheath, which causes random and fast time-varying amplitude attenuation and phase fluctuation in received signals. This paper comprehensively analyzes the mechanism of the amplitude attenuation effects on a traditional phase-locked loop (PLL), which is always ignored in traditional scenarios (such as satellite telemetry and vehicle communication). Simulation results and theoretical analysis showed that traditional PLL does not work reliably for signal carrier tracking with the severe time-varying amplitude attenuation of the plasma sheath channel. In this paper, an unconventionally designed Kalman filter (KF) tracking loop that is aware of phase dynamics and amplitude attenuation fluctuation for hypersonic vehicles is proposed. To introduce time-varying amplitude attenuation into the proposed KF-based tracking loop, the amplitude attenuation is first modeled with an autoregressive model. The statistical characteristics of the amplitude and phase fluctuation are then incorporated into the state equation and observation equation. Simulation results indicate that the proposed tracking loop is stable when the signal-to-noise ratio is −10 dB with the Ka band, even in the most severe flight environment for hypersonic vehicles.

Deep learning has become a key research component of space target detection. To ensure the safety of the aerospace field with the rapid development of hypersonic vehicles, the deep learning method is applied to recognize the flight state and type of hypersonic targets (HTs) for the first time in this article. It makes two main contributions. First, a method for constructing HT datasets based on the space-based low-orbit hyperspectral detector is proposed, which uses the nonequilibrium high-temperature flow field generated during the flight of HTs to calculate the spectral radiation characteristics and simulate the intraclass variation of each subclass of hyperspectral targets in the actual observation scene. Second, an in-depth study on the performance of deep learning methods using the constructed HT dataset is conducted, as well as a comparison with a recognition benchmark using machine learning and deep learning methods is discussed. Finally, we summarize the challenges faced by HT recognition.