This document discusses various image features that can be used for large-scale visual search and content-based image retrieval (CBIR). It describes both high-level semantic features and low-level visual features that can be extracted from images. For low-level features, it outlines several popular global features like color histograms, color moments, texture descriptors using gray-level co-occurrence matrices (GLCM), shape context, and GIST. It also discusses commonly used local feature detectors like Harris corner detector, SIFT, and descriptors like SIFT, SURF, BRIEF.