siftfeature
所属分类:图形图像处理
开发工具:matlab
文件大小:6KB
下载次数:41
上传日期:2010-09-03 11:39:51
上 传 者:
wangjian.83
说明: sift特征匹配,主要是针对目标图像和待配准图像间存在着些许变化,从而采用了尺度等不变的特征来对图像进行匹配。
(scal invariant feature transform)
文件列表:
siftfeature\it_features_matching.m (3500, 2005-05-02)
siftfeature\circ_corr_2d.m (1535, 2005-05-02)
siftfeature\features_matching.m (3663, 2005-05-02)
siftfeature\circ_corr_1d.m (1463, 2005-05-02)
siftfeature\temp_temp_matching.m (3318, 2005-05-02)
siftfeature\decimage.m (141, 2004-05-04)
siftfeature (0, 2010-09-03)
The function produce a set of features matched through a pair of views.
The selection is driven by the estimate uncertainty extracted from the Fisher information matrix.
The main script if features_matching.m
Usage:
[x1,x2,corr_vec]=features_matching(im1, im2, cal, initsplit, sm_sigma, sigma_max)
Input
im1, im2 : stereo pair. greyscale images
cal : (1=rectified images with epipolar horizontal lines; 0=uncalibrated images )
initsplit : images initial splitting factor; the images are splitted in (initsplit x initsplit) subimages.
sm_sigma : smoothing Gaussian filter SD (1 -:- 3 for normal smoothing)
sigma_max : Fisher index th (1e-2 -:- 1e-4)
Output
x1, x2 : matched points in the left and right images; 2 x N matrices
x1=[u1 u2 ....uN horiz. coord.
v1 v2 ....vN] vert. coord.
corr_vec : match strength
Comments to:
Lorenzo Sorgi
Virtual Reality lab
Italian Aerospace Research Centre
l.sorgi@cira.it
近期下载者:
相关文件:
收藏者: