Distinctive feature selection for efficient near-duplicate retrieval

The creation of distinctive subset of the features for near-duplicate image detection is significant in two terms. The former is that the query time decreases reasonably. The latter is that using the distinctive subset of the features for near-duplicate detection performs better than the ordinary subsets. In this thesis, we propose a novel interest point selection method. Using such a method, the distinctive subset is created with a density map, which is calculated from the features to rank them. We examined some experiments to show the performance of the proposed method. The recall rate of 93.08% is obtained when the precision is 98.54% in average.