The invention discloses an anisometric zooming method of video images without perceptive distortion of important content, which comprise the following steps of: firstly, packing key frames separately; then, zooming the video in an individual package, orderly defining video energy, determining the lowest energy line, selecting the optimal lowest energy line, and duplicating or deleting the same optimal lowest energy line for isometric zooming; and finally, conducting continuity processing between adjacent individual packages. The anisometric zooming method of video images without perceptive distortion of important content preprocesses the video, defines an energy function in a spatial domain and a time domain, ensures that the visual important content has high energy and the secondary content has low energy, and then deletes or duplicates the optimal energy line of a video sequence so as to complete the anisometric zooming of the video.