文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > TWO-DIMENSIONAL ENTROPY METHOD BASED ON GENETIC ALGORITHM
侵权投诉

TWO-DIMENSIONAL ENTROPY METHOD BASED ON GENETIC ALGORITHM

Abstract-Image segmentation represents the first step in many applications in image understanding. Thresholding is a simple and important in image segmentation. One-dimensional entropy does not take into account the spatial correlation between the pixels i

thresholding might give a better insight into these problems.

Abutaleb [6]extends the method of Kapur et al. using two-dimensional entropies. The two-dimensional histogram entropies are obtained from the two-dimensional histogram that is determined by using the gray value of the pixel and the local average gray value of the pixel.

Let the average gray-level values of each pixel’s neighborhood are from 0 to

L 1, as well as the gray level

values of each pixel. At each pixel, both the average gray-level value of the neighborhood and its gray level are calculated. This forms a pair that belongs to a 2-dimensional bin: the pixel gray level and the average gray level of the neighborhood. Let

fij be the total number of occurrence (frequency) of a pair (i,j), and the joint

pij is defined as

probability mass function (PMF)

pij=

fij

i,j=0,1,...,L 1.

, then the total area under

If the threshold is located at the pair (s,t)

pij(i=1,2,...,sand

j=1,2,...,t) must equal one. Thus, a normalization process is needed and this

results in a modified entropy for object

s

t

O, H(O), 1= stPst

s

t

H(O)= ∑∑

i=0j=0

pij

st

ln

pij

∑∑(p

i=0j=0

st

ij

lnpij pijlnPst)=lnPst+

Hst

st

,

where

Pst=∑∑pij,Hst= ∑∑pijlnpij

i=0j=0

i=0j=0

st

.

In a similar manner, an entropy for background

L 1L 1

B, H(B), is defined as

ln

)1 Psti=s+1j=t+1

(H Hst=ln(1 Pst)+L 1L 1

1 Pst).

H(B)= ∑∑

pijpij

1=

1 Pst)1 Pst

i=s+1j=t+1

∑∑[p

L 1L 1

ij

lnpij pijln(1 Pst)

]

The expression for H(B) is valid as long as

pij≈0

in the two regions defined by and

(i=s+1,...,L 1

and

j=1,2,...,t)

and by

(i=1,2,...,sj=t+1,...,L 1)

. This

assumption is reasonable since in many situations the off-diagonal probabilities are of negligible value. The assumption made will reduce the computation time.

The entropy-based function ψ(s,t) is defined as the sum of the two entropies, viz.,

上一页第3页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email