多尺度分形维的星载舰船显著性检测

发布时间：2018-06-29 09:48

本文选题：差分盒 + 分形维　；参考：《中国图象图形学报》2017年10期

【摘要】：目的星上的舰船检测需要在资源和时间受限条件下实现快速检测,并且对目标的种类和尺寸缺少先验信息的指导,更多时候还需要实现一景图像中不同尺寸舰船的检测,因此,星上舰船检测要求检测方法具有一定的自适应性,从而实现星上多变的检测场景。方法针对这一问题,提出了一种多尺度分形维的检测方法,可以实现一景遥感图像中不同尺寸舰船目标的检测。首先,针对差分盒算法受盒子尺寸约束的限制使分形维数的计算精度受到影响的问题提出了一种改进算法,改进算法增加了拟合直线的点对数目并引入了拟合误差剔除误差点对,提高了分形维特征计算的精确度。结果在提高了分形维计算精度的基础上,新算法利用自然物体在不同尺度上具有的自相似性,通过多尺度分形维的计算并借鉴视觉显著性中c-s算子来排除背景对目标的干扰,突出舰船目标。实验结果表明,新算法能够有效检测出一景图像中不同尺寸的舰船,优于双参数CFAR算法的检测结果。结论本文提出的多尺度分形维的检测算法可以实现对一景图像中不同尺寸舰船目标的检测,在保证一定检测率的同时有效降低了目标检测的虚警率。
[Abstract]:The target ship detection needs to realize fast detection under the condition of limited resources and time, and lacks the guidance of prior information on the type and size of the target. More often, it is also necessary to realize the detection of ships of different sizes in a scene image, so, The detection method of ship on board has certain adaptability to realize the changeable detection scene on board. Methods to solve this problem, a multi-scale fractal dimension detection method is proposed, which can be used to detect ship targets of different sizes in a remote sensing image. First of all, an improved algorithm is proposed to solve the problem that the accuracy of fractal dimension is affected by the restriction of box size on the difference box algorithm. The improved algorithm increases the number of point pairs of fitting lines and introduces fitting errors to eliminate error point pairs, which improves the accuracy of fractal dimension feature calculation. Results on the basis of improving the precision of fractal dimension calculation, the new algorithm uses the self-similarity of natural objects on different scales, and eliminates the interference of background to target by calculating multi-scale fractal dimension and using c-s operator in visual salience for reference. Highlight the ship's target. The experimental results show that the new algorithm can effectively detect ships of different sizes in a scene image, which is better than that of the two-parameter CFAR algorithm. Conclusion the multi-scale fractal dimension detection algorithm proposed in this paper can be used to detect ship targets of different sizes in a scene image, which can effectively reduce the false alarm rate of target detection while ensuring a certain detection rate.
【作者单位】：北京空间飞行器总体设计部;中国空间技术研究院通信卫星事业部;
【基金】：国家自然科学基金项目(91438102)~~
【分类号】：TP751

【相似文献】