The Atomic Force Microscope (AFM) is a powerful tool that has had a tremendous impact on the understanding of systems with nanometer-scale features. Efforts in improving the temporal resolution of the instrument have recently yielded high-speed AFMs with frame rates of approximately 10 frames per second. These instruments, however, achieve such speeds through various trade offs and as a result have limited imaging modes and scan sizes. In addition, despite these advances, typical commercial instruments continue to have frame rates well below one frame per second.This work develops and implements a novel sensing matrix for the application of compressed sensing (CS) to image acquisition in AFM, with the goal of improving the temporal resolution of the instrument by reducing the amount of data that needs to be acquired to create a high-quality image. In traditional CS, each measurement is, by design, a linear combination of the elements of the signal under study. In AFM however, the physics of the sensing process require that each measurement contains information about only a single point. The CS measurement matrix used here takes this into account and allows the user to balance image acquisition time against image quality. The proposed method is demonstrated through simulation. These simulations show faithful recovery with a reduction in imaging on the order of a factor of ten. By accepting a reduction in image reconstruction quality, additional gains in imaging time, up to a factor of twenty, were achieved.