A multi-layer approach for camera-based complex map image retrieval and spotting system

In this paper, we present a method of camera-based document image retrieval for heterogeneous-content documents using different types of features from different layers of information. We use two kinds of features in this paper (Locally Likely Arrangement Hashing - LLAH - and SIFT reduced dimensions using PCA). Then, a single hash table method is used for indexing these multiple kinds of feature vectors. In addition, we employ a technique for reducing the memory required for indexing the key points in hash table. Experimental results show that the multilayer hashing gives a high accuracy and outperforms classical methods on single layer.