Simulated annealing in sonar track detection

Abstract

Classical edge detectors cannot handle the problem of noisy sonar track detection because they are too sensitive to noise. This thesis uses Simulated Annealing for the solution by formulating the sonar track detection problem as an optimization problem. the experimental results are much better than those of classical edge detectors. Fairly good results can be obtained on data with SNR down to -18 dB where most classical edge detectors fail.