An efficient bitmap indexing method for similarity search in high dimensional multimedia databases

초록

This papers proposes a new indexing mechanism for the similarity search in high-dimensional multimedia database that quickly filter-out the irrelevant objects using bitmap index, in which the characteristic of each object is approximated as a bit-string. The bits in a bit-string that set to '1' denote the representative dimensions of object that their attribute values are relatively larger value than others. Since two objects are dissimilar if their representative dimensions are different so much, the degree of dissimilarity could be computed easily by XORing the bit-strings of two objects and counting the number of '1's in the resulting bit-string. Experimental results with more than 100,000 images show that a remarkable speed-up could be obtained with the proposed indexing method compared to VA-file and linear scan because of the simple XORing operation in the filtering process, although there are some losses in the search accuracy.