基于高斯云变换的遥感图像多粒度聚类

发布时间：2018-04-09 02:36

本文选题：遥感图像　切入点：高斯云变换　出处：《计算机科学》2017年09期

【摘要】：遥感图像技术的迅猛发展,使得传统聚类方法的局限性日益凸显。针对其信息量大、结构复杂等特点,从多粒度、多层次的角度来分析与理解地学现象,能够更好地解决遥感图像的自适应聚类问题。基于云模型与混合高斯相结合的高斯云变换是一种求解多粒度问题的新方法,能够解决问题域中多粒度的生成问题,但是其时间复杂度较高以及对噪声敏感等缺点,导致对遥感图像的聚类结果不理想。因此提出一种改进的高斯云变换方法,首先通过K-Means聚类优化初始粒度的选择,其次结合幅度云综合对粒度跃升策略进行改进,然后使用一种隶属度距离进行粒度的区域划分,最终对遥感图像进行聚类。实验结果验证了所提方法的正确性和有效性。
[Abstract]:With the rapid development of remote sensing image technology, the limitation of traditional clustering method is becoming more and more obvious.In view of the characteristics of large amount of information and complex structure, the problem of adaptive clustering of remote sensing images can be better solved by analyzing and understanding geoscientific phenomena from the point of view of multi-granularity and multi-level.Gao Si cloud transformation based on cloud model and mixed Gao Si is a new method to solve multi-granularity problem. It can solve the problem of multi-granularity generation in the problem domain, but its time complexity is high and it is sensitive to noise.The clustering results of remote sensing images are not satisfactory.Therefore, an improved Gao Si cloud transformation method is proposed. Firstly, the initial granularity selection is optimized by K-Means clustering; secondly, the granularity leap strategy is improved by combining amplitude cloud synthesis, and then a membership distance is used to partition the granularity region.Finally, the remote sensing images are clustered.The experimental results show that the proposed method is correct and effective.
【作者单位】：重庆邮电大学计算智能重庆市重点实验室;
【基金】：国家自然科学基金(61272060)资助
【分类号】：TP751

【相似文献】