注册 | 登录

SURF feature extraction on a dense grid with MATLAB

itPublisher 分享于




注:此次实现中用到的函数需要在matlab2013后的版本才能正常运行,故顺手附上matlab2014b下载地址: close all;

I would like to use SURF (speed-up robust) features to implement a classification system based on a bag-of-visual-words approach. I have read some papers about the use of a dense grid to extract these features, but I can't find how to use it with MATLAB.

Does anybody know how to do SURF feature extraction on a dense grid with MATLAB?

matlab surf
  this question
edited Mar 27 '16 at 6:48 user2227152 asked Jun 25 '13 at 11:35 mad 573 12 35 3   In matlab, surf is a kind of plot, so don't mix them up! Also, did you try googling surf feature extraction matlab? There are loads of hits. Did you try any of them? –  Hugh Nolan Jun 25 '13 at 12:45      my question is about surf (speed-up robust) features not surf (surface) plot. Matlab has a built-in function of surf feature extraction but it doesn't have an option to extract it on a grid. I also googled about other kinds of feature extraction and i found a SIFT feature extraction code on a grid in C++ which i can call using Matlab. But I want to know if there is a way to do the same with surf. Thanks for posting. –  mad Jun 25 '13 at 12:57 3   Matlab's built-in surf feature extraction is done on an grayscale image, which is a 2d data structure with presumed equal grid spacing. You can think of it as a grid. How is your data structured - a matrix or in vectors? –  Hugh Nolan Jun 25 '13 at 13:00      My image is a grayscale image and the data is in Matrix form. Thanks again. –  mad Jun 25 '13 at 13:05


1 Answers

[nRows, nCols] = size(grayImg);
STEP = 10;
colInd = (1 : STEP : nCols)';
rowInd = (1 : STEP : nRows)';
[A, B] = meshgrid(colInd, rowInd);
densePoints = [A(:) B(:)];

[featuresDense, validPointsDense] = extractFeatures(grayImg, densePoints, 'Method', 'SURF');
figure, imshow(img)
title('Dense SURF')
hold on

  this answer
answered May 26 '15 at 19:03 Swagatika 164 1 3 14      I will try this and return soon. Thank you. –  mad May 26 '15 at 21:51



本文主要演示如何使用matlab自带的Computer Vision System Toolbox这个工具箱进行suft特征点的检测、匹配及显示。这个工具箱是matlab2012b及之后才有的一个工具








您的注册邮箱: 修改

重新发送激活邮件 进入我的邮箱