Drawbacks and Proposed Solutions for Real-time Processing on Existing State-of-the-art Locality Sensitive Hashing Techniques

摘要

Nearest-neighbor query processing is a fundamental operation for many image retrievalapplications. Often, images are stored and represented by high-dimensional vectors that aregenerated by feature-extraction algorithms. Since tree-based index structures are shown to beineffective for high dimensional processing due to the well-known “Curse of Dimensionality”,approximate nearest neighbor techniques are used for faster query processing. LocalitySensitive Hashing (LSH) is a very popular and efficient approximate nearest neighbortechnique that is known for its sublinear query processing complexity and theoreticalguarantees. Nowadays, with the emergence of technology, several diverse application domainsrequire real-time high-dimensional data storing and processing capacity. Existing LSHtechniques are not suitable to handle real-time data and queries. In this paper, we discuss thechallenges and drawbacks of existing LSH techniques for processing real-time highdimensionalimage data. Additionally, through experimental analysis, we proposeimprovements for existing state-of-the-art LSH techniques for efficient processing of highdimensionalimage data.