Imaging diagnostics (ID) are key for any fusion-grade tokamak operations. The imaging diagnostic provides vital information not only about the plasma position/shape but also about the plasma interior. The main issue with ID is that they provide line-integrated information, and this integration includes different emission profiles as well as emission features. To efficiently employ ID, it is necessary to decompose this integrated line information into its integrands. Numerically speaking, tomographic reconstructions are one of the key processes by which the decomposition and local emission profiles are traditionally recovered; however, these processes are relatively slow, require a lot of computation, and have no temporal correlations. The article proposes a singular value decomposition-based method for feature selection, which decomposes not only the imaging diagnostic data for the realization of the local emission profile but also different plasma-relevant features. A test case of divertor imaging is considered for the JET tokamak in a visible imaging band for a tangential viewing geometry. The proposed method demonstrates clear divertor images for different phases of the JET tokamak plasma.
This paper presents a method to efficiently analyze data from imaging diagnostics in tokamak fusion devices. It uses Singular Value Decomposition (SVD) to decompose the integrated line-of-sight information into local emission profiles and plasma-relevant features, without the need for slow tomographic reconstructions.