In this paper we study the effects of systematic noise and atmospheric turbulence on image quality and signal-to-noise ratio for high cadence (∼20 fps) observations, in the context of fortuitous stellar occultation surveys. We performed an analysis of the spatial distribution of the weighted centroid of a reference star, in the X and Y direction (in the detector), and we also analyze the FWHM behavior to study the effect on the stellar profile. This information allowed us to find both the shift and the deformation of each image, which are the most significant effects of atmospheric turbulence on such images. Our observations were carried out at the 84 cm telescope of the San Pedro Mártir Observatory, located in Ensenada, Baja California, México. In order to isolate the atmospheric turbulence contribution on image quality and motion, the systematic noise due to telescope motion and other low frequency components were filtered-out from the distribution curve for the centroids. We used the APPHi pipeline as our analysis tool to study the effect of the size of the Region of Interest (ROIs) on energy loss in the photometry of the stars. We find for example that using ROI of 7 × 7 pixels in our study, the shifting and deformation of the images will produce an energy loss of ∼10%, for more than 9% of the observations, even after removing the systematic noise. Such energy loss can be comparable to the flux drop expected for some TNO occultations, thus leading to spurious detections. We also find that the algorithm developed to calculate an optimal ROI in the APPHi pipeline, makes a good compromise between avoiding the flux loss and keeping a reasonably small ROI, to reduce the data storage requirements and maximize the number of stars that can be simultaneously observed while limiting the overlap of ROIs for stars which are very close to each other.