基于高分辨率遥感图像的船只纹理研究

Research of Ship Texture Based on High Resolution Remote Sensing Images

为实现长江流域有关部门对江面目标船只进行管理,及满足对特定时段、特定区域的船只类型、分布和数量进行监控的需求,开展了基于高分辨遥感图像计算机自动提取船只纹理的研究。选择了相应的统计学指标作为评价因子,利用强大的IDL作为开发工具,实现了船只纹理研究的专项功能。通过目视解译对计算结果进行验证,结果表明选择指标合理,所有纹理因子中,熵、梯度和相关性最能反映纹理特征,并存在着一定的相关性。

In this paper, a ship texture automatically extracted from high resolution remote sensing images is researched. The purpose of study is to manage some specific ships in Changjiang River according to requirements of relative departments, and monitor types, distribution and quantity of ships in a specific time and a specific region. Adopting indexes in statistics is selected as estimate factors. Application to ship texture has been realized by IDL （ interactive data language）, a powerful tool in process of images. By analyzing visual interpretation and results of experiment, it is shown that selected index is reasonable. Entropy, gradient and relativity between them, are the most appropriate texture factors to represent features of the ship texture.