The indexing process contains inevitably a step of choice of key
frames, representative of each consistent unity, extracted by both the
transition detection and the building of the hierarchy.
We will not discuss here the choice of the key frames. Once this choice is done, the following step consists in the indexing of these still images, and for this, a segmentation into large homogeneous regions corresponding to the scene object is essential. This step allows in effect to apply global operators on regions homogeneous in color or in texture for example.
For a description of the different steps of this segmentation and an illustration of its results, click here. Let us just remind one more time that the segmentation aim is only to extract large homogeneous regions, without taking small details into account.