c语言写harris角点提取难吗
发布网友
发布时间:2022-04-24 00:46
我来回答
共1个回答
热心网友
时间:2023-11-02 13:26
针对
Har
ris
角点检
测算法对
T
型和斜
T
型
角点存在定
位不准确
以及运算速
度慢的问题
,
提出
了一种改
进算法。
改进算法计算目标像素点的
8
邻域范围内与之灰度相似的点的
数目
,
然后对目标
像素点周
围其他像素
点按同
样方法
处
理
,
并对得到的计算数据进行比较
,
分析出局部范围内的像素
点的灰度值
分布。根据比
较结果
,
从中遴
选出部
分像素
点
作为下一步角点检测的计算对象
,
并计算其角
点响应函数值
,
如
果角点响应
函数值大
于设定的阈
值
,
该
点则会
被确定
为
最终的角点。实验结果表明
:
改进算法的角点检测时间
仅为原
算法的
14.
3%
,
并实现
了对
T
、
斜
T
型角
点准确
定位
,
提
高了
H
arr
is
角点检测算法的检测效率和精度。
关
键
词
:
机器视觉
;
图像处理
;
角点检测
;
H
arr
is
算法
;
角点响应函数
中图分类号
:
T
P391.
4
文献标识码
:
A
An
improved
algorithm
for
Harris
corner
detection
WANG
Wei
1,
2
,
TANG
Yi
ping
1
,
REN
Juan
li
3
,
SH
I
Bing
chuan
1
,
LI
Pei
lin
2
,
H
AN
H
ua
ting
2
(
1
.
State
K
ey
L
abor
atory
f
or
Manuf
acturing
Sy
stem
Eng
ineer
ing
,
X
i
an
J
iaotong
University
,
X
i
an
710049
,
China;
2
.
L
aunch
E
ngineer
ing
Dep
ar
tm
ent,
M
issile
I
nstitute
of
Air
Force
E
ngineering
Univer
sity
,
Sany
uan
713800
,
China;
3
.
S
chool
of
A
rt
and
Media
,
X
i
an
T
echnological
Univer
sity
,
X
i
an
710032
,
China)
Abstract:
An
impr
oved
algo
rithm
for
H
arris
cor
ner
detectio
n
is
proposed
by
considering
the
lim
itatio
ns
of
inaccurate
localization
to
tw
o
kinds
of
corners
(
T
and
diag
onal
T
)
and
lo
w
efficiency
in
H
ar
ris
algo
r
ithm.
T
he
im
pro
ved
alg
orithm
is
used
to
calculate
the
number
o
f
points
that
the
g
ray
level
is
similar
to
that
of
the
target
pixel
w
ithin
8
neighborhoods
of
the
tar
get
pixel,
and
then
to
tr
eat
the
other
points
in
the
sam
e
manner.
Som
e
of
the
pixels
are
selected
as
the
calculating
objects
for
the
nex
t
step
of
cor
ner
detection
by
analy
zing
distr
ibution
of
the
pixels
in
lo
cal
area
and
com
paring
the
calculated
data,
and
the
v
alues
for
their
corner
response
functions
are
calculated.
if
the
value
is
larg
er
than
threshold,
it
w
ill
be
co
nsidered
as
the
final
corner.
T
he
experimental
results
show
that
the
com
putation
tim
e
of
impr
oved
algorithm
is
o
nly
14.
3%
that
of
the
or
ig
inal
H
arris
alg
orithm,
and
it
can
realize
the
accur
ate
localization
to
T
and
diag
onal
T
corners,
w
hich
m
eans
that
im
pro
ved
algo
rithm
acquires
better
eff