Imagery analysis

Abstract: The U.S. Navy has plans to develop an automated system to analyze satellite imagery aboard its ships at sea. Lack of time for training, in combination with frequent personnel rotations, precludes the building of extensive imagery interpretation expertise by shipboard personnel. A preliminary design starts from pixel data from which clouds are classified. An image segmentation is performed to assemble and isolate cloud groups, which are then identified (e.g., as a cold front) using neural networks. A combination of neural networks and expert systems is subsequently used to transform key information about the identified cloud patterns as inputs to an expert system that provides sensible weather information, the ultimate objective of the imagery analysis.

Abstract: : A new class of remote sensing data with great potential for the accurate identification of surface materials is termed hyperspectral imagery. Airborne or satellite imaging spectrometers record reflected solar or emissive thermal electromagnetic energy in hundreds of contiguous narrow spectral bands. The substantial dimensionality and unique character of hyperspectral imagery require techniques which differ substantially from traditional imagery analysis. One such approach is offered by a signal processing 'paradigm, which seeks to detect signals in the presence of noise and multiple interfering signals. This study reviews existing hyperspectral imagery analysis techniques from a signal processing perspective and arranges them in a contextual hierarchy. It focuses on a large subset of analysis techniques based on linear transform and subspace projection theory, a well established part of signal processing. Four broad families of linear transformation-based analysis techniques are specified by the amounts of available a priori scene information. Strengths and weaknesses of each technique are developed. In general, the spectral angle mapper (SAM) and the orthogonal subspace projection (OSP) techniques gave the best results and highest signal-to-clutter ratios (SCRs). In the case of minority targets, where a small number of target pixels occurred over the entire scene, the low probability of detection (LPD) technique performed well.

Abstract: Synthetic Aperture Radar (SAR) imagery has been widely used in many maritime applications due to its high resolution, wide coverage, and real-time monitoring characteristics. Nevertheless, the size of SAR images is significantly large for real-time application. In recent years, High-Performance Computing (HPC)-related methods have been used to improve the precision and detection rate of SAR imagery analysis. In this paper, motivated by the state-of-the-art real time object detection You Only Look Once version 3 (YOLOv3), an enhanced GPU-based deep learning method has been proposed, namely Accelereated-YOLOv3 (A-YOLOv3), to detect ships from the SAR images. A-YOLOv3 aims to reduce the computational time with relatively competitive detection accuracy by constructing a new architecture with less layers and channels. The proposed A-YOLOv3 architecture achieves Average Precision (AP) of 97.4% on the Expand Diversified SAR Ship Detection Dataset (EDSSDD).

Abstract: The invention features a system wherein a recognition environment utilizes comparative advantages of automated feature signature analysis and human perception to form a synergistic data and information processing system for scene structure modeling and testing, object extraction, object linking, and event/activity detection using multi-source sensor data and imagery in both static and time-varying formats. The scene structure and modeling and testing utilizes quantifiable and implementable human language key words. The invention implements real-time terrain categorization and situational awareness plus a dynamic ground control point selection and evaluation system in a Virtual Transverse Mercator (VTM) geogridded Equi-Distance system (ES) environment. The system can be applied to video imagery to define and detect objects/features, events and activity. By adapting the video imagery analysis technology to multi-source data, the invention performs multi-source data fusion without registering them using geospatial ground control points.