A Reliability-Based Multi-Algorithm Fusion Technique in Detecting Changes in Land Cover

Remote Sensing

Penglin Zhang¹  Wenzhong Shi¹  Man Sing Wong¹ 

[1] School of Remote Sensing and Information Engineering, Wuhan University, No. 129 of Luoyu Road, Wuhan 430079, China; E-Mail:

关键词: reliability;    change-detection;    multi-algorithm fusion;    fuzzy technique;   

DOI  :  10.3390/rs5031134

来源: mdpi

【 摘 要 】

Detecting land use or land cover changes is a challenging problem in analyzing images. Change-detection plays a fundamental role in most of land use or cover monitoring systems using remote-sensing techniques. The reliability of individual automatic change-detection algorithms is currently below operating requirements when considering the intrinsic uncertainty of a change-detection algorithm and the complexity of detecting changes in remote-sensing images. In particular, most of these algorithms are only suited for a specific image data source, study area and research purpose. Only a number of comprehensive change-detection methods that consider the reliability of the algorithm in different implementation situations have been reported. This study attempts to explore the advantages of combining several typical change-detection algorithms. This combination is specifically designed for a highly reliable change-detection task. Specifically, a fusion approach based on reliability is proposed for an exclusive land use or land cover change-detection. First, the reliability of each candidate algorithm is evaluated. Then, a fuzzy comprehensive evaluation is used to generate a reliable change-detection approach. This evaluation is a transformation between a one-way evaluation matrix and a weight vector computed using the reliability of each candidate algorithm. Experimental results reveal that the advantages of combining these distinct change-detection techniques are evident.

【 授权许可】

CC BY   
© 2013 by the authors; licensee MDPI, Basel, Switzerland.