<p>Fast pressure-sensitive paint (FPSP) allows for measuring the pressure distribution in a variety of challenging conditions. However, a key issue is processing the recorded intensity data to allow accurate reconstruction of the pressure field. Most challenging is accurately filtering out noise from the measured intensity field, while preserving the underlying measured pressure distribution. A common approach is a local filtering operation, such as Gaussian-weighted pixel averaging. Alternatively, a global approach can be taken, where the entire measured intensity field is mapped via a least squares (LS) minimization routine. In this paper, we introduce a regularized LS-approach called RELS, which further allows filtering of noise. Therefore, this paper explores the effects of either local Gaussian filtering or global LS or RELS data filtering on the measured pressure field across the surface. The results demonstrate that local filters are more time-efficient than global filters when the number of images to process is low. However, when the number of images to process is significant, global approaches are faster, since the global interpolation matrix only needs to be constructed once. Moreover, global approaches achieve lower error bounds compared to the Gaussian filter at similar filter size, and are recommended even at the cost of slower computation time. For the RELS approach, the standard deviation of the identified pressure field decreases by a factor of two compared to a Gaussian filter of similar size.</p>