Targets detection of sonar images is a branch of underwater targets detection, and is important for the analysis of Synthetic Aperture Sonar (SAS) images. Nowadays, the most used detection method of sonar images is based on intensity of target echoes, with some digital image processing method. But the effect is not very impressive. However, when applied to SAS images, the detection results are improved if the properties of the specific images are considered.

So LI Changzhi, TIAN Jie, ZHANG Yangfan, HUANG Haining and ZHANG Chunhua of Institute of Acoustics, Chinese Academy of Sciences took a series of studies and proposed an automatic targets detection method from SAS images by analyzing the statistical properties of the images.

The method mainly has three procedures: preprocessing of SAS images, especially the mean-standard deviation segmentation method to binarize the images, extracting areas of connected components in the binary images as features, maximum area of connected components method detecting. Finally, the researchers carried out the practical experiment and the result showed that the targets detection method based on mean-standard deviation was validated on lake-trial datasets.