遥感图像中复杂海面背景下的海上舰船检测

Ship Detection in Remote Sensing Images with Complex Sea Surface Background

针对遥感图像中复杂海面背景下海上舰船的检测问题展开讨论,在Itti视觉显著度模型的基础之上进行改进,提出一种基于特征显著度图的复杂海面上舰船的自动检测方法,解决了传统的阈值分割方法在遥感图像复杂海面背景下较难将目标与背景分离的问题.在多种不同复杂海面背景下的舰船检测实验中,与传统阈值分割方法比较,本文方法有较高的检测率和较低的虚警率.

In this paper, the problem of ship detection in remote sensing images with complex sea surface background is discussed. Traditional segmentation approaches have much difficulty in separating targets from complex sea surface background. To overcome the deficiency, the Itti visual saliency model is introduced and improved by a global saliency measurement, and an automatic approach based on saliency map is proposed for ship detection in remote sensing images with complex sea surface background. The results of different experiments indicate that this approach is robust and effective. Comparing with a traditional approach based on threshold segmentation, the correct ratio of this approach is increased and the missed alarms ratio is reduced.