Multi-resolution segmentation parameters optimization and evaluation for VHR remote sensing image based on meanNSQI and discrepancy measure

ABSTRACT

Multi-Resolution Segmentation (MRS) is known to be a general segmentation algorithm for very-high-resolution (VHR) remote sensing applications. The critical problems of MRS are the optimization of the parameters and the evaluation of segmentation quality. Based on the principle of maximizing the intra-object homogeneity and inter-object heterogeneity, we propose a novel Normalized Segmentation Quality Index (NSQI) and use level filtering to acquire the optimal parameters of the MRS algorithm. Using the geometric and arithmetic discrepancy between the segmented object and the reference object as the evaluation criterion, we then evaluate the quality of the segmented objects. The results of two experiments confirm the effectiveness of our meanNSQI and discrepancy measure approach. Additionally, a sensitivity analysis of the segmentation quality index demonstrates the reliability of the meanNSQI and the discrepancy measure.

KEYWORDS:

• Multi-resolution segmentation
• discrepancy measure
• very high resolution remote sensing
• parameters optimization

Acknowledgments

The authors gratefully acknowledge the Digital Globe Corporation for providing Worldview-2/3 remotely sensed images. Also, we thank Stuart Jenkinson, PhD, from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the The Fundamental Research Funds of Beijing University of Civil Engineering and Architecture [X18055]; Beijing Advanced Innovation Center for Future Urban Design [UDC201650100]; National Natural Science Foundation of China [41771412,41801235]