Ambient noise imaging(ANI)derives Green's functions that characterize subsurface medium properties from seismic noise recorded at stations, thereby enabling imaging of the Earth's internal structure. Unlike conventional seismic methods that rely on earthquake or artificial sources, ANI offers superior spatiotemporal resolution. However, existing ANI theory generally assumes uniformly distributed noise sources, neglecting their intrinsic spatiotemporal heterogeneity. Mismatches between this idealized assumption and real-world conditions can systematically bias imaging results. This review synthesizes current understanding of ambient noise source characteristics and their spatiotemporal variability, summarizes two decades of progress in identifying source-induced biases in cross-correlation functions and ANI outcomes, and discusses key unresolved challenges in the field.