PROJECT TITLE :

Scalable Feature Matching by Dual Cascaded Scalar Quantization for Image Retrieval

ABSTRACT:

In this paper, we tend to investigate the matter of scalable visual feature matching in large-scale image search and propose a unique cascaded scalar quantization theme in twin resolution. We tend to formulate the visual feature matching as a range-based neighbor search problem and approach it by identifying hyper-cubes with a twin-resolution scalar quantization strategy. Specifically, for every dimension of the PCA-remodeled feature, scalar quantization is performed at each coarse and fine resolutions. The scalar quantization results at the coarse resolution are cascaded over multiple dimensions to index an image database. The scalar quantization results over multiple dimensions at the fine resolution are concatenated into a binary super-vector and stored into the index list for efficient verification. The proposed cascaded scalar quantization (CSQ) technique is freed from the expensive visual codebook training and therefore is independent of any image descriptor training set. The index structure of the CSQ is flexible enough to accommodate new image options and scalable to index massive-scale image database. We have a tendency to evaluate our approach on the general public benchmark datasets for large-scale image retrieval. Experimental results demonstrate the competitive retrieval performance of the proposed method compared with several recent retrieval algorithms on feature quantization.