The Fourier-transforming properties of coherent imaging systems and their applications in spatial filtering techniques are well known and thoroughly developed. Most of these previous systems were static and unable to adapt for inputs exhibiting different sets of characteristics. I present an adaptive coherent imaging system utilizing simulated annealing in the frequency domain. The iterative process of simulated annealing enables the filter to adapt to a variety of inputs and automatically optimize even in the presence of random noise. A set of possible guidelines for parameter selection is presented to facilitate the implementation of this technique into similar applications.