An adaptive simulated annealing cooling schedule for object detection in images

Description

In our image processing applications, we use a simulated annealing procedure to find configurations of geometric shapes that fit the best an image. This type of algorithm allows finding one of the global minima of an arbitrary function provided that the cooling schedule is logarithmic with the time. Since this type of cooling schedules is very slow, geometrical cooling schemes are used in practice. Geometrical schemes are however subject to some disadvantages that we discuss in this report. To overcome these disadvantages, we propose an adaptive cooling scheme. This heuristic is based on the analysis of the cooling scheme behavior in practice. In particular, we observe the presence of critical temperatures. To deal with these critical temperatures, we propose a cooling scheme that decelerates when such a temperature is detected, and accelerates otherwise. We present results on a real problem taken from our image processing applications.

Additional details