秀域集团怎么样:好难，好郁闷．谁能帮我翻译一下

3. Texture Algorithms
A texture map, generated from a fractal dimension algorithm is expected to
produce a measure that is the same value as the fractal dimension. If that
measure were capable of capturing all aspects of the real texture and, at the
same time, the texture-generating algorithm based on that measure was a
capable algorithm, the other textural metrics (correlation length, etc.) would
also be expected to provide the same measure for the real and synthetic
textures. If some aspect of this expectation is not met, then further study will
be suggested. It is not expected that any one of these algorithms will be
successful 100 percent of the time. Therefore, we chose three measures to
use in three different algorithms to generate texture.
The mid-point displacement algorithm is a method of generating a twodimensional
fractional Brownian motion or noise. This approximate method is
described in The Science of Fractal Images. [9] The mid-point displacement
algorithm produces an image that only approximates a fractional Brownian
"motion". Another algorithm, based on this same concept, does a better job.
[lo] See appendix A.
The GLCM method for building the textures presented here, is based on the
algorithm described in the paper by G. Lohmann. [l I] In the algorithm, a set
of four grey-level co-occurrence matrices provide feature vectors. Lohmann
reports that the four primary co-occurrence matrices 并发矩阵
(horizontal, vertical, and
left- and right-diagonal) contain sufficient information to synthesize texture
that closely resemble textures from remotely-sensed images of the Landsat/TM
and ERS-1IAM satellite-borne sensors. Our idea is to determine how well this
algorithm performs on our type of imagery.
For a description of the correlation length algorithm used in this study, see
appendix G where the basic methodology is laid out: [12]
Numerous approaches to texturing were reviewed and a
simplified two-dimensional, autoregressive (AR) model was
selected. It uses correlation length in vertical and horizontal
directions and brightness mean and standard deviation as input
parameters for a kernel (process of order [1,1]). The model was
developed from an AR model used in the U.S. Air Force Infixed
Modeling and Analysis (IRMA) image modeling system.
An empirical approach was taken because of the lack of general
theory or models on thermal IR texture. Using this approach
homogeneous surfaces (grass, bare soil, trees, tree lines, etc.) of
interest were imaged for the times (or under similar conditions)
金山快译我也会用啊