A linear filter that correlates data with a proposed signal, weighting frequencies according to their noise. In stationary Gaussian noise it maximizes the expected signal-to-noise ratio for that known signal.
For data , known complex source template , real amplitude , and known noise covariance , generalized least squares gives the displayed amplitude. This convention takes in the quadratic objective ; factors of two in the amplitude variance depend on whether complex-component or real-component covariance is used. Source position, shape and contaminants must be fitted or marginalized when uncertain. An extended-source complex visibility template accounts for resolved-out flux rather than assuming the source is unresolved.
Articles by others on the same topic
A matched filter is a signal processing technique used to maximize the signal-to-noise ratio (SNR) when detecting a known signal in the presence of noise. It is particularly useful in situations where the signal is embedded in a noisy environment, such as in radar, communications, and image processing. ### Key Concepts 1. **Purpose**: The primary purpose of a matched filter is to identify the presence of a specific signal shape (or template) within a received noisy signal.