ABSTRACT
This chapter shows that the changepoint detection theory has clear practical applications for the development of efficient algorithms that are both easily implemented and have certain optimality properties. It addresses the problem that is motivated by certain multisensor target track management applications in active sonar systems. The chapter discusses the problem of in-frame detection of streaks of faint space objects with unknown orbits in CCD frames. The problem is solved for streaks with a very low pixel signal-to-noise ratio of the order of 1 and smaller. An effective solution of this problem can be obtained on the basis of joint testing of hypotheses and estimation of unknown parameters (beginning and end of the streak) by maximizing the likelihood ratio, i.e., by the generalized likelihood ratio method.
