Atomic Force Microscopy (AFM) is a powerful tool that can perform nano-scale imaging. Normally AFM tip is controlled to scan on sample surface line by line to get the topographic image and this process takes several minutes. Higher sample rate is demanded so that when doing continuous imaging, the time interval between each image can be significantly shorted thus video-imaging can be achieved. In this paper, a compressive sensing based AFM video-imaging system is built, and random walk based scanning path is proposed. Compressive sensing requires random sampling. Bio-inspired random walk based scan path is able to provide a random tip moving path, which enables compressive sensing to be implemented into AFM scanning system. Experiments based on this system are set up in order to test the performance. It first raster scan the entire area and then generate a biased random tip moving path focusing on some specific areas. Compressive scan is then used to continuously scan the sample surface. Finally, video-imaging is achieved and dynamic changes in nano-scale are observed.