It was hard to discriminate inshore ships from harbors because of the similarity of gray and texture for both targets.Aiming at these problems,a method of harbor ships target detection based on secondary seg-mentation was proposed.Firstly,a top-hat operator was used to watershed segmentation algorithm to solve the over-segmentation phenomenon,and obtain the image of sea and land separation results at the same time.Sec-ondly,make use of Freeman Chain Code tracking the harbor ships contour and combine with the improved Hough conversion to detect the target to separate the dock and ship through the secondary segmentation.Finally calculate the regional area to achieve the harbor ships recognition.The experiment results show that this method has strong robustness and easy to implement,experiments based on the satellite sensing image demonstrated that the method proposed could solve the harbor ships detection problem.%针对光学遥感图像中港口舰船的纹理、灰度特征与码头十分接近，直接分割相对困难的问题，提出了基于二次分割的港口舰船目标检测方法。该方法首先将 Top-hat 算子引入分水岭分割算法中，解决了分水岭算法过分割现象，同时获得较完整的海陆分离结果；其次对分割后的陆地区域使用 Freeman 链码进行轮廓描述，并结合改进的 Hough变换进行目标区域的分离，通过二次分割有效地将码头与舰船分开；最后，计算各区域的面积实现港口舰船的识别。实验结果表明，该方法鲁棒性强，易于实现，通过实测数据可以解决港口舰船目标的检测问题。
展开▼
