An Object Appearance Detection in a Sequence of Noisy Images using Stochastic Approximation

Y. Pyatkin and G. Salov (Russia)


Noisy image sequence, object detection, cumulative sum test, stochastic approximation.


In the first part of the present report we consider the mathematical problem of adaptive approximation of the unknown likelihood ratio (likelihood functional) for detection of the important object on a single noisy image. In the second part this approximation of the likelihood functional is applied to the problem of quickest detection of the time appearance of the important object in an observed sequence of independent noisy images. A fast detection algorithm based on cumulative sums test is presented. An example is discussed.

Important Links:

Go Back