This study aims to improve automated target detection in Bearing Time Records (BTRs) images, addressing key challenges such as low signal-to-noise ratios and false alarms. Motivated by the need for more reliable detection methods in marine environments, we propose three key techniques: (1) Global–Local Peak initialization to optimize cluster center setup, (2) an adaptive method using SSE derivatives for precise determination of cluster count, and (3) a fuzzy rule based on correlation coefficient histograms to reduce false alarms. The proposed approach demonstrates 100% accuracy in simulations and has proven highly effective in sea trials, significantly enhancing the reliability of target detection in marine settings.
Improvements of bearing time records target extraction method based on culstering
