The present invention discloses a mid-air gesture input method and apparatus. In embodiments of the present invention, when a writing start command is detected, gesture images are collected, and a position of a fingertip in each frame of the gesture images is acquired; a writing trajectory is generated according to the acquired positions of the fingertip in the gesture images; and when a writing end command is detected, text recognition is performed on the generated writing trajectory, to obtain text corresponding to the writing trajectory. In the embodiments of the present invention, gesture images of a user are collected to generate a writing trajectory, and text recognition is then performed on the writing trajectory, so that the user only needs to use a gesture in mid-air to implement text input, and the input manner is simple and applicable to many electronic products including a television.