Automatic time-series quantification of bluff erosion using a single consumer grade camera as basis for erosion risk assessment and forecasts – a Boston Harbor Islands case study

详细信息    查看全文

• 作者：Martin D. Hellwig
• 关键词：Computer vision ; Geomorphology ; Time series ; Boston Harbor Islands ; Citizen science
• 刊名：Journal of Coastal Conservation
• 出版年：2016
• 出版时间：December 2016
• 年：2016
• 卷：20
• 期：6
• 页码：469-476
• 全文大小：
• 刊物类别：Earth and Environmental Science
• 刊物主题：Geography, general; Coastal Sciences; Oceanography; Nature Conservation; Remote Sensing/Photogrammetry;
• 出版者：Springer Netherlands
• ISSN：1874-7841
• 卷排序：20

文摘

Many communities along coastlines and riverbanks are threatened by water erosion and hence an accurate model to predict erosion events is needed in order to plan mitigation strategies. Such models need to rely on readily available meteorological data that may or may not be correlated with the occurrence of erosion events. Computing these correlations requires a quantified index that reports the magnitude of erosion events over time. This study introduces a method to create erosion indices using affordable consumer grade digital cameras. It is able to detect and quantify erosion using an image series obtained from just one such camera by segmenting each images instance into equally sized squares that can be preprocessed and analyzed separately. This approach isolates each image segment from noise and temporary disturbances that frequently occur throughout images taken with low cost cameras. In this fashion, noise may either be addressed locally or simply ignored if it is too extreme. After preprocessing, comparison of subsequent segments yields change matrices that form the basis for segment-specific erosion indices that are later combined into a composite value for the entire image instance. Whenever a segment instance is unavailable or unusable, the algorithm attempts to use neighboring instances whenever possible. When tested against human observation of erosion events during a 6 week period, the resulting index achieves a true positive rate of 67% while producing only a small number of false positives. Finally, the index is validated by significant correlation with various meteorological data streams.