aum

Auburn University at Montgomery computer science professor awarded NSF grant for cloud detection research

Auburn University at Montgomery (AUM) Assistant Professor of Computer Science Semih Dinc was awarded a $328,252 National Science Foundation (NSF) grant to develop a computer-based tool that identifies cloud regions using sky images in real time.

The three-year NSF grant study will be led by Dinc, principal investigator, and co-led by Luis Cueva Parra, assistant professor of computer science at the University of North Georgia. Randy Russell, assistant professor of astronomy at AUM, will serve in a senior role on the project.

“The goal of this project is to develop an open-source software framework to identify cloud regions using technology that can achieve a higher accuracy image of clouds in real-time,” Dinc said. “Our framework could be used by other researchers to solve problems in the fields of weather prediction, climatology, agriculture, and hydrology.”

Several existing studies suggest that clouds are changing due to climate change and indicate at the same time clouds affect global warming, Dinc said. His team’s research will assist with detecting the amount of solar light and radiation reaching the ground, which is directly related to the density of clouds, he said.

Assistant Professor Semih Dinc

In August, Dinc and his research team began working to develop the learning-based machine using technology known as Hyperspectral Imagery (HSI) capable of such detection.

“Existing cloud detection research mainly utilizes conventional color cameras, and mostly they do not have a real-time processing goal,” Dinc said. “HSI has a significantly higher spectral resolution when compared to conventional camera images. A pixel on the HSI can be represented by hundreds of features, which are collected from a wider range of the electromagnetic spectrum.”

Because of the vast amount of information HSI technology can process, his team also will employ High Performance Computing (HPC) tools, such as cluster computers and general purpose (GPU) accelerators, to achieve the real-time performance needed for more accurate cloud detection, Dinc said.

“For us, cloud detection is a challenging problem from a mathematical and computational point of view,” he said. “HSI has never been tried before for this purpose so it’s important to provide current literature that offers a reliable and accurate system that runs in real-time for use by researchers and industries that rely on cloud data.”

Dinc’s NSF grant will provide for 27 undergraduate research assistantships at AUM over the three-year project’s duration.

“The study will directly impact our undergraduate students as they will be exposed to today’s important topics such as machine learning, computer vision, and high-performance computing,” he said.