/Fast-Match

Fast matching algorithm for sift feature points

Primary LanguagePython

Fast matching algorithm for sift feature points

Abstract

Today’s cameras produce images that often exceed 10 megapix els. Yet computing and matching local features for images of this size can easily take 20 seconds or more using optimized matching algorithms. This is much too slow for interactive applications and much too expensive for large scale image operations. We introduce Fast-Match, an algorithm designed to match large images efficiently without compromising matching accuracy. It derives its speed from only computing features in those parts of the image that can be confidently matched. Fast-Match is an order of magnitude faster than the popular Ratio-Match, yet often doubles matching precision for difficult image pairs.

Paper and Attribution

You can find the paper presenting this work here: Fast Match.

If you make use of this code or ideas expressed in the paper, please cite the following paper:

J. T. Arnfred, S. Winkler.
Fast-Match: Fast and robust feature matching on large images. 
Proc. IEEE International Conference on Image Processing (ICIP), Québec City, Canada, Sept. 27-30, 2015.

How to install

Just git clone it

Code example

Run the following code in the main directory

import cv2
from cache import Metric_Cache, Grid_Cache
import fastmatch
import figures

# Open images and create cache
target_path = "images/graf/img1.ppm"
query_path = "images/graf/img4.ppm"
query_cache = Metric_Cache(query_path)
target_img = cv2.imread(target_path)
query_img = cv2.imread(query_path)

# Match using Fast-Match with a ratio threshold of 0.7
log = []
match_fun = fastmatch.match(query_cache, target_img, options = {'log' : log})
matches = list(match_fun(0.7))

# Visualize results
# The [:,:,::-1] inverses the color channels so the images are printed correctly
figures.visualize_log(log, query_img[:,:,::-1], target_img[:,:,::-1], stop_at = 100)

License

The code is released under the MIT license.

If you make use of this code, please cite the Fast-Match paper at ICIP 2015.