A method of indexing a video document represents the video document in the form of a parallelepipedal volume obtained by juxtaposing, on a time axis, successive frames of the video document where each frame is represented by a matrix of pixels in a two-dimensional space. A space-time video slice of the video document is constructed by extracting a set of pixels for each frame of the video document in a same line defined relative to a first direction of the frames and juxtaposing the lines of pixels corresponding to the frames on a time axis to form a space-time video slice of a length defined by the juxtaposed lines of pixels and of a height equal to the length of the lines extracted from each of the frames. The length is a function of the dimension of the frames in a second direction and the orientation of the lines. The colorimetric values of the pixels of the lines of the space-time video slice are initialized on the basis of the content of the lines extracted from the frames of the video document.