Background Noise Cancellation from Active Sonar Images Utilizing Transmission-free Recordings

Sonar images are generally exposed to various types of noise. In this paper, we focus on self-noise reduction of sonar systems and propose an efficient approach by applying region-wise thresholds acquired by a statistical analysis of transmission-free recordings from the same sonar. With this approach in many cases up to 95 % of noise pixels can be eliminated while maintaining 98 % of signal pixels at an extremely low computational effort.