Learning-AI

Review of Image Descriptors

November 2019

This review is based on the pyimagesearch course.

Image descriptors and feature vectors quantify and abstractly represent an image using only a list of numbers.

Feature vector: An abstraction of an image used to characterize and numerically quantify the contents of an image. Normally real, integer, or binary valued. Simply put, a feature vector is a list of numbers used to represent an image.
- this feature can be passed down to image classifier or image search engine
Image descriptor: An image descriptor is an algorithm and methodology that governs how an input image is globally quantified and returns a feature vector abstractly representing the image contents.
- they tend to be much simpler than feature descriptors
- HoG, LBPs, Harlick texture
Feature descriptor: A feature descriptor is an algorithm and methodology that governs how an input region of an image is locally quantified. A feature descriptor accepts a single input image and returns multiple feature vectors.
- SIFT, SURF, ORB, BRISK, BRIEF, and FREAK

Keypoint detection and feature extraction:

keypoints are simply the (x, y)-coordinates of the interesting, salient regions of an image.
feature extraction is the process of extracting multiple feature vectors, one for each keypoint How to use multiple features per image? Keypoint matching or bag-of-visual-words.

Fast algorithm to detect corners
Fusing Points and Lines for High Performance Tracking ICCV 2005
A test is performed for a feature at a pixel p by examining a circle of 16 pixels (a Bresenham circle of radius 3) surrounding p. A feature is detected at p if the intensities of at least 12 contiguous pixels are all above or all below the intensity of p by some threshold, t.
openCV example