Imaging system range defines the maximal distance at which a selected object can be seen and perceived following surveillance task perception criteria. Thermal imagers play a key role in long-range surveillance systems due to the ability to form images during the day or night and in adverse weather conditions. The thermal imager range depends on imager design parameters, scene and transmission path properties. Imager range prediction is supported by theoretical models that provide the ability to check range performance, compare range performances for different systems, extend range prediction in field conditions, and support laboratory measurements related to range. A condensed review of the theoretical model’s genesis and capabilities is presented. We applied model-based performance calculation for several thermal imagers used in our long-range surveillance systems and compared the results with laboratory performance measurement results with the intention of providing the range prediction in selected field conditions. The key objective of the paper is to provide users with reliable data regarding expectations during a field mission.

No abstract

For the last two decades a large number of different automatic modulation classification (AMC) algorithms were developed, and many improvements in classification performance are reported. This was commonly achieved by engaging complex structures of neural networks, or other adaptable mechanisms for achieving better precision, when it comes to decisionmaking. Still, from practical implementation point of view, low algorithm complexity, economical usage of resources and fast execution remain to represent very desirable properties of an AMC algorithm. These properties are recognized in AMC algorithms based on higher-order cumulants as classification features, so their further improvement is of interest. Previous performance analysis of an algorithm based on sixthorder cumulants, in scenarios with complex valued signals' classification, showed that improvements are possible in the context of resources engaged and speed of execution. In this paper a novel approach is presented, for improving the correctness of classification process with sixthorder cumulants and simple twostep feature extraction structure, by engaging a new method for reduction of observed signal's modulation order which directly improves the classification performance. While tested with sixthorder cumulants, proposed method preserves good statistical properties of signal's higher-order cumulants in general, so it can be adopted in other AMC algorithms as well. Proposed modulation order reduction method is described in details, tested through computer simulations within the sixthorder cumulant AMC algorithm, and achieved improvements in performance are presented and explained.

No abstract

SWIR imaging bears considerable advantages over visible-light (color) and thermal images in certain challenging propagation conditions. Thus, the SWIR imaging channel is frequently used in multi-spectral imaging systems (MSIS) for long-range surveillance in combination with color and thermal imaging to improve the probability of correct operation in various day, night and climate conditions. Integration of deep-learning (DL)-based real-time object detection in MSIS enables an increase in efficient utilization for complex long-range surveillance solutions such as border or critical assets control. Unfortunately, a lack of datasets for DL-based object detection models training for the SWIR channel limits their performance. To overcome this, by using the MSIS setting we propose a new cross-spectral automatic data annotation methodology for SWIR channel training dataset creation, in which the visible-light channel provides a source for detecting object types and bounding boxes which are then transformed to the SWIR channel. A mathematical image transformation that overcomes differences between the SWIR and color channel and their image distortion effects for various magnifications are explained in detail. With the proposed cross-spectral methodology, the goal of the paper is to improve object detection in SWIR images captured in challenging outdoor scenes. Experimental tests for two object types (cars and persons) using a state-of-the-art YOLOX model demonstrate that retraining with the proposed automatic cross-spectrally created SWIR image dataset significantly improves average detection precision. We achieved excellent improvements in detection performance in various variants of the YOLOX model (nano, tiny and x).