Title: An image-plane algorithm for JWST's non-redundant aperture mask data

Abstract: The high angular resolution technique of non-redundant masking (NRM) or aperture masking interferometry (AMI) has yielded images of faint protoplanetary companions of nearby stars from the ground. AMI on James Webb Space Telescope (JWST)'s Near Infrared Imager and Slitless Spectrograph (NIRISS) has a lower thermal background than ground-based facilites and does not suffer from atmospheric instability. NIRISS AMI images are likely to have 90 - 95% Strehl ratio between 2.77 and 4.8 micron. In this paper we quantify factors that limit the raw point source contrast of JWST NRM. We develop an analytic model of the NRM point spread function which includes different optical path delays (pistons) between mask holes and fit the model parameters with image plane data. It enables a straightforward way to exclude bad pixels, is suited to limited fields of view, and can incorporate effects such as intra-pixel sensitivity variations. We simulate various sources of noise to estimate their effect on the standard deviation of closure phase, sigma_CP (a proxy for binary point source contrast). If sigma_CP < 10^-4 radians --- a contrast ratio of 10 magnitudes --- young accreting gas giant planets (e.g. in the nearby Taurus star-forming region) could be imaged with JWST NIRISS. We show the feasibility of using NIRISS' NRM with the sub-Nyquist sampled F277W, which would enable some exoplanet chemistry characterization. In the presence of small piston errors, the dominant sources of closure phase error (depending on pixel sampling, and filter bandwidth) are flat field errors and unmodeled variations in intra-pixel sensitivity. The in-flight stability of NIRISS will determine how well these errors can be calibrated by observing a point source. Our results help develop efficient observing strategies for space-based NRM.