Interactive Visualization of Three-Dimensional Descriptors Using Virtual Reality

Techniques of information visualization aims at finding visual representation of data sets in order to increase their understanding, allowing better analysis. Content-Based Image Retrieval (CBIR) allows to retrieve images from a database using image features extracted by descriptors, which are algorithms that represents numerically the images. When CBIR is applied to three-dimensional (3D) image domain, information understanding is more difficult due to the intrinsic abstract nature of the information. Interaction in real-time in 3D virtual environment provided by Virtual Reality techniques can help the understanding because it makes possible to analyze data from different point of views. This paper aims to define and implement visualization techniques for data generated by two three-dimensional descriptors developed for using in 3D CBIR systems. The results showed that the visualization techniques contribute to validate extractors both for convex and non-convex models.