czf_blurcue
所属分类：matlab编程
开发工具：matlab
文件大小：40KB
下载次数：152
上传日期：2011-11-17 12:01:08
上 传 者：
ishwaryaj
说明： In this paper, a multimodal image fusion algorithm based on multiresolution
transform and particle swarm optimization (PSO) is proposed.
Firstly, the source images are decomposed into low-frequency coefficients and
high-frequency coefficients by the dual-tree complex wavelet transform
(DTCWT). Then, the high-frequency coefficients are fused by the maximum selection
fusion rule. The low-frequency coefficients are fused by weighted average
method based on regions, and the weights are estimated by the PSO to gain
optimal fused images. Finally, the fused image is reconstructed by the inverse
DTCWT. The experiments demonstrate that the proposed image fusion method
can illustrate better performance than the methods based on the DTCWT, the
support value transform (SVT), and the nonsubsampled contourlet transform
(NSCT).
文件列表:
czf_blurcue (0, 2010-04-04)
czf_blurcue\biker.jpg (34300, 2010-04-03)
czf_blurcue\initstft.m (967, 2010-04-04)
czf_blurcue\dostft.m (773, 2010-04-04)
czf_blurcue\compsni.m (654, 2010-04-04)
czf_blurcue\compski.m (612, 2010-04-04)
czf_blurcue\demo.m (1282, 2010-04-04)
czf_blurcue\conv2r.m (799, 2010-04-04)
czf_blurcue\invldmap.m (1119, 2010-04-04)
czf_blurcue\pyk.m (1069, 2010-04-03)
This MATLAB source code provides an implementation of the local blur
cue described in the following paper:
Ayan Chakrabarti, Todd Zickler and William T. Freeman, "Analyzing
Spatially-varying Blur", in Proc. of the IEEE Computer Society
Conference on Computer Vision and Pattern Recognition (CVPR), 2010.
This code is (as of now) released only for non-commercial research
use. Please cite the above paper if you use this code to generate any
results included in an academic publication.
This code currently can be used to analyze the likelihood of 1-D
kernels working at different locations in the image, and was used as a
part of the whole segmentation algorithm (to be released separately)
described in the paper.
Please see demo.m for an example implementation of how to use the
provided functions. Broadly,
- initstft -- Initializes filters for sub-band representation for a
particular window size.
- compski -- Computes the local spectrum of any kernel under the
representation.
- compsni -- Computes the noise in different bands of gradients
of the image.
- dostft -- Actually computes magnitudes of the sub-band coefficients
at all locations in the image.
- pyk -- Computes, for every window in the image, the log likelihood of
a particular kernel being active in that window.
- invldmap -- Helper function to compute where coefficients / likelihood
values are unreliable because the window contained pixels that were
saturated / clipped, or outside the image boundary.
Please e-mail questions/comments to: ayanc@eecs.harvard.edu
近期下载者：
相关文件：
收藏者：