Bayesian Cloud Extraction for Assessment of Space-Debris Impact Using Conditional Entropy

Abstract

High-speed collisions of space debris with space structures produce debris clouds, the properties of which determine the collision resistance required of space structures. Debris clouds impact the pressurized walls of space structures; therefore, it is important to determine the contours of the debris clouds to support an efficient protective design of space structures. This paper proposes a new Bayesian cloud contour extraction method for accurately extracting the contours of debris clouds from images obtained in high-speed collision experiments. The method employs conditional entropy as an indicator of the extraction accuracy. In a departure from the conventional assumption of a Gaussian distribution, a realistic probability distribution is proposed to describe the histograms of image data. A new and more versatile method for cloud contour extraction is also proposed based on the trisection of the histograms. Using collision experiment images, it is demonstrated that the two proposed methods are superior to the conventional methods in terms of accurately extracting the debris cloud contours. The proposed methods can be used to measure the shape of a debris cloud and its rate of expansion, which will improve the interpretation of images of impact and explosion phenomena and contribute to the development of methods to protect space structures from space debris.