基于统计量的声呐图像目标检测算法

水下声呐图像目标检测问题是一项重要而困难的工作,采用滑动窗计算图像中各像素点处邻域像素灰度的统计量,利用最大熵图像分割算法确定检测阈值,并利用均值、标准差、偏态和峰度等统计量对算法进行了仿真验证,对声呐图像中的目标回波区和阴影区域均可实现较好的检测效果。结果表明,该方法具有原理简单、运算效率高、实时性好等特点,具有较强的工程应用价值。

Algorithm of target detection in sonar imagery based on statistical value

The target detection of underwater sonar imagery is a crucial task. The statistical value of pixels grey-level based on sliding-windows was calculated, certain entropy criterion was used to gain the detecting threshold. Mean,standard deviation, skewness and kurosis were respectively adopted to validate the performance of arithmetic. The target highlights region and shadow region of sonar image can be detected precisely. Simulation results indicate that this method presented in this paper has characteristics of simple principle, high efficiency and real time, then it has certain value for engineering application.