Subspace Clustering for Information Retrieval in Urban Scene Databases

We present a comprehensive study of two important subspace clustering algorithms and their contribution to enhance results for the difficult task of matching images of the same object using different devices at different conditions. Our experiments were performed on two distinct databases containing urban scenes which were tested using state-of-the-art matching algorithms. Our start point was the hypothesis that low discriminant local point descriptors lead to misclassification, which can be reduced employing clustering techniques as filters. A significantly amelioration of the results obtained for the two tested databases was achieved, which indicates that subspace clustering techniques have much to contribute at this research area. Another point is whether the occurrence of obstacles like trees and shadows are responsible for misclassification of images.