2026-05-19T03:35:58Zhttps://keep.lib.asu.edu/oai/request

oai:keep.lib.asu.edu:node-2013802025-05-21T21:15:50Zoai_pmh:alloai_pmh:repo_items

201380 https://hdl.handle.net/2286/R.2.N.201380 http://rightsstatements.org/vocab/InC/1.0/ http://creativecommons.org/licenses/by-nc-sa/4.0 2025-05 61 pages Edwards, Amelia Cochran, Douglas Zhou, Shuang Barrett, The Honors College School of Mathematical and Statistical Sciences In this study, we examine an approach for detection of changes and motion estimation that is applicable to pixelated images of sea ice, both optical (photographic) and Synthetic Aperture Radar (SAR). This approach is based on the magnitude-squared coherence (MSC) estimate, a statistic that is well studied in connection with testing whether two noisy time series contain a common but unknown signal component. Following a description of our data, we proceed to derive the distribution of the MSC estimate formed from a pair of time series that are real-valued under explicit null hypotheses. We proceed to investigate the relevance of this null hypothesis to model segments of sea ice imagery that are known to not be related by the phenomena of interest. This assessment involves transformation of sea ice imagery by local normalization to approximately fit the null hypothesis used in our theoretical derivation. We then describe an algorithm that quantifies and detects both motion approximation and change within a pair or series of images. The algorithm is demonstrated for both applications using synthetic data, real sea ice imagery, and hybrid data in which patches of synthetic imagery are inserted into real sea ice images to provide examples where change and motion are precisely characterized. Sea Ice Mathematics Statistics Statistical Distributions Change Detection Registration Coherence-Based Change Detection for Sea Ice Imagery