主动声呐B显图像中的疑似目标自动拾取算法

Automatic picking up algorithm of suspected targets in the B-type image of active sonar

针对主动声呐B显图像呈“满天星”状从而导致人工提取目标困难的问题,本文在分析B显图像干扰分布特点和界定疑似目标概念的基础上,综合运用图像分割、图像形态学、统计拟合与检测等技术,提出一种疑似目标自动拾取算法。研讨了图像自动调亮与二值化、形态学闭运算与前景背景标定、双阈值检测与综合筛判等算法关键环节的设计与实现,并使用海试数据对算法性能进行了检验。结果表明:本文算法能自动区分B显图像的前景和背景区,从中稳定可靠地自动拾取疑似目标,漏报率低。

To address the problems of the B-display image of active sonar being "all over the sky" and of the difficulty in picking up the targets manually, an automatic algorithm for picking up suspected targets is proposed based on the analysis of the interference distribution characteristics of B-type image and the definition of the concept of suspected targets by comprehensively applying image segmentation, image morphology, statistical fitting, and detection. The design and implementation of key links of the algorithm, such as automatic image brightening and binarization, morphological closed operation, foreground and background calibration, double threshold detection, and comprehensive screening, are discussed. The performance of the algorithm is tested by using sea trial data. Results show that this algorithm can automatically distinguish the foreground area from the background area of the B-type images and automatically pick up the suspected targets stably and reliably with a low level of miss alarm rate.