A level set-based framework for typhoon segmentation with application to multi-channel satellite cloud images

This paper presents a segmentation framework for delineate typhoon to improve the accuracy of typhoon forecast. An algorithm of automatically locating ROI of typhoon without interactive selection is designed. The C-V model is extended to the vector space and an energy term without re-initialization is introduced. An iteration convergence condition is proposed for avoiding unnecessary iterations. Through the use of multichannel cloud image information and excluding the unnecessary iterative calculation, this framework can accelerate the speed of curve evolution and improve the accuracy of segmentation. The experimental results verify the validity and feasibility of the proposed method and achieve the expected typhoon segmentation effect.