simpsal
所属分类:matlab编程
开发工具:matlab
文件大小:100KB
下载次数:343
上传日期:2011-03-22 20:36:56
上 传 者:
endchaos
说明: 基于视觉注意机制的图像显著图提取,包括亮度,dkl颜色,方向等通道,合成图像的显著性描述图
(Image-based visual attention saliency map mechanism extraction, including brightness, dkl color and direction of channels, the significance of synthetic image description graph)
文件列表:
simpsal\simpsal\attenuateBordersGBVS.m (1148, 2010-03-18)
simpsal\simpsal\default_fast_param.m (2806, 2010-02-20)
simpsal\simpsal\default_pami_param.m (275, 2010-02-20)
simpsal\simpsal\getchan.asv (2741, 2010-12-08)
simpsal\simpsal\getchan.m (2741, 2010-12-08)
simpsal\simpsal\gist_sal.asv (3320, 2010-12-14)
simpsal\simpsal\lena.jpg (69214, 2010-02-19)
simpsal\simpsal\mexLocalMaximaGBVS.cc (1785, 2010-02-19)
simpsal\simpsal\mexLocalMaximaGBVS.dll (6656, 2010-12-07)
simpsal\simpsal\mexLocalMaximaGBVS.mexa64 (7539, 2010-02-19)
simpsal\simpsal\mexLocalMaximaGBVS.mexglx (5866, 2010-02-19)
simpsal\simpsal\mexLocalMaximaGBVS.mexmaci (8952, 2010-02-19)
simpsal\simpsal\mexLocalMaximaGBVS.mexmaci64 (8872, 2010-03-05)
simpsal\simpsal\mexLocalMaximaGBVS.mexw32 (6656, 2010-02-19)
simpsal\simpsal\mexLocalMaximaGBVS.mexw64 (8192, 2010-02-19)
simpsal\simpsal\myconv2.m (294, 2010-02-19)
simpsal\simpsal\mygausskernel.m (218, 2010-02-19)
simpsal\simpsal\mynorm.m (117, 2010-02-19)
simpsal\simpsal\mypeakiness.m (161, 2010-02-19)
simpsal\simpsal\my_simpsal_demo.m (523, 2010-02-19)
simpsal\simpsal\padImage.m (797, 2010-02-19)
simpsal\simpsal\pixsal.m (334, 2010-02-20)
simpsal\simpsal\rgb2dkl.m (9700, 2010-02-20)
simpsal\simpsal\shiftImage.m (453, 2010-02-19)
simpsal\simpsal\simpleGabor.asv (852, 2010-12-07)
simpsal\simpsal\simpleGabor.m (852, 2010-12-07)
simpsal\simpsal\simpsal.asv (2778, 2010-12-08)
simpsal\simpsal\simpsal.m (2808, 2010-12-08)
simpsal\simpsal (0, 2011-01-05)
simpsal (0, 2011-01-05)
simpsal.zip
Saliency (MATLAB source code)
http://www.klab.caltech.edu/~harel/share/gbvs.php
by Jonathan Harel
jonharel@gmail.com
California Institute of Technology
If you use this code, please cite it as:
Jonathan Harel, A Saliency Implementation in MATLAB: http://www.klab.caltech.edu/~harel/share/gbvs.php
================================================================================
This is an installation and help file for the saliency map (MATLAB) code here.
What you can do with this code: Compute the standard Itti, Koch, Niebur (PAMI 19***)
saliency map for an image or frame sequence (i.e. video), and simplications thereof.
This is intended to be a very simple implementation so that it can be easily
modified.
================================================================================
Fast start-up procedure:
(1) Put this directory somewhere on your disk, then add the directory to your
MATLAB path, or change into this directory to call the code here.
(2) Try running my_simpsal_demo
open my_simpsal_demo.m to see how it works.
Basically, at the matlab prompt you can call:
>> map = simpsal(img);
To get a saliency map 'map' on an input image 'img'. To have it resized
to the resolution of the input image, use this:
>> mapbig = mat2gray(imresize( map , [ size(img,1) size(img,2) ] ));
================================================================================
Helpful Notes:
(1) the first argument to simpsal() can be an image name or image array
* if the image array is 4 dimensional ( H x W x {1 or 3} x T ), it is possible
to compute temporal feature channels including Flicker and Motion.
One way to use this on a video would be to pass in a frame buffer
of the last ~15 frames and take map(:,:,end) as the saliency map
for the current frame.
(2) there is an optional, second, parameters argument
* default_pami_param is used if there is no second parameters argument.
That tries to approximate the original Itti Algorithm as closely as possible.
* default_fast_param is a setting of parameters which allows you to
compute saliency maps very quickly. It excludes an orientation channel,
and map weighting according to peakiness.
(3) See default_fast_param.m to see what the parameters do.
(4) See getchan.m to understand how features are computed.
===============================================================================
Problems / Errors / Troubleshooting :
1. Binaries are included for Windows, Linux, and Mac OS X intel.
However, if you run into problems running mexLocalMaximaGBVS, you may have
to compile mexLocalMaximaGBVS.cc by typing mex('mexLocalMaximaGBVS.cc')
at the matlab prompt when you are in this directory. If you have never
compiled mex files before, you probably have to run 'mex -setup' first.
2. If you run into problems running prctile, mat2gray, imshow, or imresize.
Download http://www.klab.caltech.edu/~harel/share/stupidtools.zip
And put the contents in this directory.
================================================================================
Copyright Notice:
parts of mexLocalMaxima.cc taken from
http://ilab.usc.edu/toolkit/
================================================================================
Revision History
first authored 12/7/2009
Revised 2/19/2008 fixed bug with Gabor filters
Revised 3/18/2010 added attenuateBordersGBVS to orientation call
近期下载者:
相关文件:
收藏者: