针对海岸线背景下的海面小目标的自动检测问题展开讨论,提出一种目标所在感兴趣区域的自动提取算法,将水域和复杂背景分离,并对分割后的二值图像进行标记,海面区域即水域应该是标记块中面积最大的一块,利用此特征提取出感兴趣区域--水域.对感兴趣区域进行轮廓跟踪后再用原始灰度进行扫描填充,在得到的简单背景下对目标进行局部中值滤波,进而用F-R准则将感兴趣舰船目标检测出来.最后拟利用感兴趣舰船目标区域的方差值计算感兴趣目标出现的置信度估计值,若置信度大于90%,则认为是真目标.实验结果给出了实验处理时间和所识别出的各个目标的置信度,表明了本文算法的有效性.

A new method is presented to detect small target in the port-sea background image.The image contains land and water area,and the target is in the water area, so the algorithm of automatic segmenting the area that we are interested in is presented.Firstly the land and water area are divided,then the black-white image which has been segmented is marked and the biggest block is distilled as the water area.Secondly the area＇s contour is scouted and it is filled with the original pixels. A local median median filter is executed to filter the rest bundance background elements.Eventually the real small target can be segmented by the F-R rule.And the experimental result shows that the method is effective