Gas cloud infrared image enhancement based on anisotropic diffusion

Leakage of dangerous gases will not only pollute the environment, but also seriously threat public safety. Thermal infrared imaging has been proved to be an efficient method to qualitatively detect the gas leakage. But some problems are remained, especially when monitoring the leakage in a passive way. For example, the signal is weak and the edge of gas cloud in the infrared image is not obvious enough. However, we notice some important characteristics of the gas plume and therefore propose a gas cloud infrared image enhancement method based on anisotropic diffusion. As the gas plume presents a large gas cloud in the image and the gray value is even inside the cloud, strong forward diffusion will be used to reduce the noise and to expand the range of the gas cloud. Frames subtraction and K-means cluttering pop out the gas cloud area. Forward-and-Backward diffusion is to protect background details. Additionally, the best iteration times and the time step parameters are researched. Results show that the gas cloud can be marked correctly and enhanced by black or false color, and so potentially increase the possibility of gas leakage detection.