Visual Exploration of 3D Shape Databases via Feature Selection

We present a visual analytics approach for constructing effective visual representations of 3D shape databases as projections of multidimensional feature vectors extracted from their shapes. We present several methods to construct effective projections in which different-class shapes are well separated from each other. First, we propose a greedy heuristic for searching for near-optimal projections in the space of feature combinations. Next, we show how human insight can improve the quality of the constructed projections by iteratively identifying and selecting a small subset features that are responsible for characterizing different classes. Our methods allow users to construct high-quality projections with low effort, to explain these projections in terms of the contribution of different features, and to identify both useful features and features that work adversely
for the separation task. We demonstrate our approach on a real-world 3D shape database.