A Multilevel Visual Feature-Based Approach for Measuring the Spatial Information in Remote Sensing Images

Abstract

The information measurement of remote sensing images plays an important role in various geographical applications. As people are the ultimate interpreter of remote sensing images, there is a need to integrate principles of human visual perception into remote sensing image information measurement. Therefore, this article proposes an approach to measure the spatial information of remote sensing images on the basis of visual features. The information measurement model suitable for remote sensing images is first developed according to the Shannon–Hartley theorem. Then, the multidimensional remote sensing image information measurement strategy is proposed based on the multilevel visual features. On this basis, the remote sensing image information can be quantified by incorporating the low-level visual features. Experiments demonstrate that the proposed method can capture the information content of remote sensing images that is highly consistent with human visual cognition. Through comparisons with other methods, it is demonstrated that the proposed method has advantages in accurately revealing the attention rule for the visual information of remote sensing images.