Optical Flow Methods for PUNCH Imaging

Abstract:

Optical flow is one of the fundamental techniques in the field of Computer Vision. Optical flow is the distribution of apparent velocities of movement of brightness pattern in an image. Horn & Schunck 1981 defined a general algorithm for determining optical flow using a simple regularization term to constrain the problem. Implementations such as Lucas & Kanade 1981 and processing techniques such as course -to-fine pyramids have been incorporated to improve the robustness of the technique. Optical flow has been applied to LASCO C2 coronal data. However, past applications were limited by the cadence of the observations relative to the scene velocity and the signal to noise ratio. Beyond cadence and image quality, coronal observations have unique challenges due to the diffuse and dynamic nature of space plasma not found in other applications. New observations coming from PUNCH with higher cadence and signal to noise have renewed interest in these powerful tools. When applied to PUNCH images, optical flow would provide large scale solar wind velocities and velocity map of CME in an automated way. We will explore different optical flow techniques to determine which is most promising for the new observations from PUNCH.