顾及纹理特征贡献度的变化影像对象提取算法

遥感影像变化检测是全球变化研究的重要内容。基于两期遥感影像的变化检测方法存在数据条件要求苛刻、难以充分利用快速发展的多源遥感影像数据等问题。目前许多变化检测的参考数据中包含了一期分类矢量数据,矢量数据中往往包含了位置、形状、大小和类别属性等先验信息,充分利用这些先验信息将可提高变化检测精度。提取变化影像对象是结合矢量数据和遥感影像进行变化检测的核心步骤。本文提出了一种顾及纹理特征贡献度的变化影像对象提取方法。该方法利用矢量数据分割遥感影像,获取影像对象,计算影像对象纹理特征值。根据信息增益原理计算纹理特征参数的特征贡献度,选择特征参数。由贡献度指数大小确定纹理特征参数权重,计算影像对象与先验要素类别的相似度系数,提取变化影像对象。试验结果表明,基于纹理特征贡献度的特征参数选择,能有效地提高变化影像对象提取结果的精度。

Changed Image Objects Extraction Algorithms Considering Texture Feature Contribution

Remote sensing image change detection is an important part of global change research.The change detection methods based on two-temporal remote sensing images consist of drawbacks which affect the accuracy of change detection results,such as rigorous data requirements,inadequate adoption of multi-source remote sensing image data.At present,there are some existing classification vector dataset available for change detection in many regions,and some prior knowledge are included in the existing classification vector dataset,e.g.,the position,shape,size and class.Making full use of the prior information is beneficial to improve the accuracy of change detection result.Extracting changed image objects is the key step in the change detection using the existing vector data and the latest remote sensing image,Therefore,a new change detection method based on texture feature contribution is proposed.The vector data is used to segment remote sensing image,the image objects can be extracted,and the texture feature value of image objects can be calculated.According to the principle of information gain,the feature contribution of texture feature parameters is defined,and it is used to select texture feature parameters for texture feature analysis.A similar coefficient of texture feature is defined and is used to extract changed image objects.The experimental results show that selecting texture feature parameters based on feature contribution can effectively improve the accuracy of extracting changed image object resu l t.