图像中的目标通常具有复杂的形状和尺寸,现有方法难以充分挖掘地物的显著性空间信息。基于此,提出一种基于显著性测度的形态学显著性剖面。首先,根据图像区域内部灰度和轮廓信息计算显著性测度,用于描述目标在场景中的重要程度,然后提取具有显著性测度局部极大值的重要区域,并通过多层级特征描述其空间信息。形态学显著性剖面的构造首先利用基于显著性测度的属性滤波滤除图像的冗余细节,并保留图像的显著结构;再根据图像中显著的组织结构生成层次化的空间特征。实验采用了两组高光谱数据集进行验证,实验结果表明所提算法的分类效果优于其他形态学特征提取算法。

Generally, the objects in an image have complex shapes and sizes. Therefore, it is difficult for the existing morphological features to completely describe the significant spatial information of the image. Hence, a morphological saliency profile is developed in this study based on the saliency measure. The grayscale and contour information of a particular area can be used to estimate the value of the saliency measure. This measure is used to describe the importance of a target in a scene. Thus, the important area of an image can be extracted based on the local maximum value of the saliency measure, and its spatial information can be obtained based on the multi-level features. When extracting the morphological saliency profile, attribute filtering based on the saliency measure is performed to eliminate redundant image details and retain the saliency profile of the image. Subsequently, the hierarchical spatial features are generated according to the saliency of the organization structure in the image. Two hyperspectral datasets are used in this experiment for verification. The experimental results demonstrate that the classification performance of the proposed algorithm is superior to those of the existing morphological feature extraction algorithms.