Author(s): Manuel Fiallos-Salguero; Soon-Thiam Khu; Jingyu Guan; Tianzhi Wang; Mingna Wang

Linked Author(s): Mingna Wang

Keywords: Computer vision algorithms; Image and video processing; Rainfall estimation; Rainfall monitoring; Spatial-temporal resolution

Abstract: Rainfall has been considered a crucial resource for effective management and control in various fields (e. g., flood control); however, the rainfall data recorded by traditional monitoring methods shows deficient accuracy and reflects low spatiotemporal resolution. Based on that, suitable rainfall measurement methods need to be developed to enhance the extreme rainfall track and reduce the vulnerability of considerable damage that might occur in urban areas. This study develops a new approach based on computer vision algorithms and enhanced geometrical optics analyses to improve rain streak detection and rainfall estimation from videos with complex backgrounds. Thus, the proposed approach revealed a satisfactory reduction in the error rate for the rainfall estimation with the enhancement of data collection, extraction, and measuring, showing promising results for rainfall monitoring applied to complex-real rain event scenarios at different times.

DOI: https://doi.org/10.3850/iahr-hic2483430201-284

Year: 2024