Two fundamental motivations exist for designing remote sensing systems that are undersampled: (1) SNR considerations that motivate larger detectors to collect more photoevents per frame time and (2) the desire to maximize the field of view with a finite number of detectors. As a result, many remote sensing systems do not satisfy the Nyquist sampling criterion, leading to measured images corrupted with a defect called aliasing. We describe a localized subpixel motion sensing algorithm that is used to properly place small blocks of the sequence of images in an upsampled space. Localized motion sensing enables dealiasing to be performed on image sequences with more complicated relative motion than simple translation. The presented results show improvements in both the edge response and subjective image quality.