Answers are grounded in this paper and cite their sources.
This paper describes the development of a target tracking algorithm for an anti-ship imaging infrared seeker. The tracking algorithm uses feature extraction and machine learning to discriminate between a desired target ship and decoys within the seeker’s field of view. The algorithm is developed within a high fidelity simulation architecture, used to simulate engagements of infrared missiles against ships, aircraft and land vehicles. The proposed seeker tracking algorithm will be evaluated in a naval engagement scenario, against a ship deploying countermeasures. The tracking algorithm performs the tasks of object detection, feature extraction and target selection. Object detection is achieved via thresholding the image of the seeker’s field of view, and thereafter, shape and intensity based features are calculated for each resulting object. These features are then used as inputs to a neural network, which performs the task of target selection, to determine the seeker’s aim-point. Object features such as peak and average intensity, intensity moments, eccentricity, roundness, minimum, maximum and average radial perimeter distances are considered, to determine their discriminatory power. A training set of images of different ships and decoys, generated by the front end of a seeker model within the simulation architecture, is used to obtain a comprehensive collection of these features. An analysis is performed to determine which of the features are the most discriminatory and these are then used as inputs to the neural network. The neural network is trained on these features to recognise the difference between ships and decoys. Examples of the performance of the tracking algorithm will also be shown.
No takes yet. Share an insight, caveat, or question.
Gray et al. (2012) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: