基于显著区域提取的红外图像舰船目标检测

针对复杂海面背景下红外图像舰船目标由于灰度不均匀、海杂波干扰大等因素造成的自动检测虚警率高、准确率低的问题，提出了一种显著区域提取和目标精确分割相结合的红外舰船目标检测方法。首先，利用基于图论的视觉显著性(Graph-based Visual Saliency ,GBVS)模型计算待检测图像的显著图，使得目标区域信息增强；其次，结合舰船目标先验信息（长短轴、面积等），利用多级阈值划分算法提取关注的显著区域，并确定原图中候选目标区域；最后，利用空间约束模糊C均值(Fuzzy C-Means,FCM)算法对候选区域进行分割，结合目标先验知识对分割区域筛选并输出目标位置。所提方法在公开数据集IRShips上与相关方法进行比较，结果表明，相比直接进行全图目标搜索的方法，所提方法不仅准确率高、执行速度快，且检测目标的位置更加精确。

Ship Detection in Infrared Image Based on Salient Region Extraction

To deal with the problem of automatic ship target detection with high false alarm rate and low accuracy in infrared image under complex background caused by the factors such as intensity inhomogeneity and sea clutters,a ship detection method based on the combination of salient region extraction and accurate target segmentation is proposed.Firstly,the saliency map of the test image is computed by graph-based visual saliency(GBVS) model,so that the information around the target can be enhanced.Secondly,the region of interest is extracted by multi-level threshold algorithm based on the prior information(length,width,area,etc.) of the ship,and the candidate target areas in test image can be determined.Finally,the candidate areas are segmented using spatial constraint based fuzzy C-means(FCM) segmentation algorithm,and the segmentation areas satisfying the prior information are regarded as the real targets.The proposed method is evaluated on IRShips dataset and the results show that the proposed framework is effective and performs better compared with the methods directly searching the target in entire image.