Going one step ahead feedback querying in integrating users into a search process, navigation is the more recent approach to finding images in a large image collection by using content-based information. Rather than using queries or going into a feedback querying process that would be both heavy in terms of human-computer interaction and computer processing time, navigation on a pre-computed data structure is easier and smoother for the user. In particular, we found Galois' lattices to be convenient structures for that purpose. However, while properties extracted from images are usually real-valued data, most of the time a navigation structure has to deal with binary links from an image (or a group of images) to another. A trivial solution to get a binary relationship from real-valued data is to apply a threshold, but this solution not only leads to a loss of information but also tends to create sparse areas in the lattice. In this paper, we propose a technique to incrementally build a Galois' lattice from real-valued properties by taking into account the existing structure, thus limiting the size of the lattice by avoiding the creation of sparse nodes. Experiments showed that this technique produces a navigation structure of better quality, making search process faster and more efficient, thus improving user's experience.