基于多特征加权的SAR影像舰船检测优化方法

针对虚警目标较多的复杂场景中,传统舰船检测算法检测结果精度偏低的问题,提出了一种基于多特征加权的SAR影像舰船检测优化方法。首先,采用标记分水岭算法对SAR幅度影像进行去陆操作;其次,利用基于对数正态分布的恒虚警率算法,得到去陆SAR影像的候选目标;再次,提取候选目标的长宽比、舰船面积和对比度3个特征;最后,提出变异系数法对3个特征进行权重分配,并结合候选目标的归一化特征矢量计算其特征置信度,再确定最佳置信度,去除候选目标中的虚警目标,优化舰船检测结果。为了验证所提方法,选取不同复杂场景的高分三号SAR影像进行舰船检测实验。实验结果表明,所提方法具有可行性和有效性。

Ship detection optimization method in SAR imagery based on multi-feature weighting

Aiming at the problem that the accuracy of traditional ship detection algorithms is not satisfying in complex scene with many false alarm targets,a ship detection optimization method in SAR imagery based on multi-feature weighting was proposed.Firstly,the marker-based watershed algorithm was employed to remove land from SAR amplitude image.Then,the CFAR algorithm based on log-normal distribution was used to obtain candidate targets from no land image.Furthermore,the length to width ratio,the ship area and the contrast ratio of the candidate targets were extracted.Finally,a variance coefficient method was proposed to distribute the weight of the three features,and the confidence levels were calculated by combining the normalized feature vectors of the candidate targets with the feature weight.By determining the best confidence level,false alarm targets among the candidate targets were removed to optimize ship detection results.In order to verify the proposed method,experiments were carried on with the GF-3 SAR images of different complex scenes.The experimental results show that the proposed method is feasible and effective.