Enhanced Anomaly Detection Algorithm For Hyperspectral Line Scanning Using CNN and ERX

Hyper spectral imaging (HSI) has emerged as a powerful remote sensing technology capable of capturing detailed spectral information across hundreds of contiguous wavelength bands.This high spectral resolution enables precise material identification and anomaly detection in diverse application domains such as environmental monitoring, defense surveillance, industrial inspection, agriculture, and mineral exploration.However, the increasing volume and dimensionality of hyperspectral data present significant computational challenges, However, existing line-scan algorithms are too slow when using small computers (e.g., those onboard a drone or small satellite) particularly in real-time line-scanning systems where rapid processing is essential.Traditional anomaly detection methods, including Support Vector Machine (SVM)-based approaches, often suffer from high computational cost and limited adaptability to streaming hyper spectral data.This research introduces ERX (Efficient Real-time RX), a fast anomaly detection algorithm specifically designed for hyper spectral line-scanning systems.This paper proposes an anomaly detection method based on the random selection of background pixels,the random-selection-based anomaly detector (RSAD) ERX is an optimized adaptation of the classical RX (Reed-Xiaoli) detector, reformulated to support incremental covariance estimation and reduced computational complexity while maintaining high detection accuracy.The proposed approach employs several convolutional and pooling layers to extract deep features from HSIs, Unlike conventional batch-processing approaches, ERX operates effectively in streaming environments by leveraging recursive statistical modeling and dimensionality optimization techniques.The proposed ERX framework addresses three primary challenges: (1) high-dimensional spectral data processing, (2) computational inefficiency in covariance matrix inversion, and (3) latency constraints in realtime anomaly detection.Through adaptive background modeling and optimized matrix computation strategies, ERX achieves significant performance improvements compared to SVM-based anomaly detection systems.Experimental evaluation demonstrates that ERX provides superior detection accuracy, reduced false alarm rates, and substantially lower processing latency, making it highly suitable for embedded and real-time hyperspectral applications.This paper introduces the Exponentially moving RX algorithm (ERX) to address these issues.This study contributes a novel real-time anomaly detection methodology that balances computational efficiency and detection reliability The results indicate that ERX is particularly effective for realtime line-scanning hyperspectral systems, offering scalable performance and improved anomaly localization capabilities.