Figure 6
An illustration of the bag of visual words approach. The first row shows the process of learning a vocabulary of visual words by computing the ORB descriptor and clustering the collection of descriptors into groups whose centres will define the visual words by K-means clustering. The second row shows how we use the visual word tree. Given a diffraction pattern, we compute the ORB descriptors. For each descriptor we then find the closest cluster centre and increment the frequency count for that visual word. The result is a histogram of visual word counts. |