Common Spatial Pattern Filtering for Detection of Circumstellar Discs

We present a new algorithm to be used for extraction of circumstellar discs from direct imagining observations. The underlying mechanism of this algorithm is Common Spatial Pattern filtering, a technique commonly found in fields outside astronomy. It is a method used to maximize the difference between two sets of data. We employ this by using distributing the disc signal in different locations, employing Angular Differential Imaging. By generating the difference, we can reconstruct the circumstellar disc in post-processing with limited speckle noise. We demonstrate the algorithm with a common disc: HR 4796a. These results are then compared to a current algorithm: Karhunen-Loeve Image Processing. Common Spatial Pattern filtering represents a new class of direct imaging signal extraction: modelling the signal directly rather than speckle subtraction.