opencv的特征匹配算法性能评估报告
需积分: 49 107 浏览量
2016-02-17
15:06:49
上传
评论 3
收藏 113KB DOCX 举报
Test Report
Test plaorm:
Hardware: Lenovo K800
System version： (factory reset)4.0.4
Net environment: No sim card，wi" o#，bluetooth o#
All tests run in screen-locked status
Console-based program started in adb.exe shell
Build environment： Win7 + eclipse
Performance
B
R
I
S
K
B
F
F
A
S
T
+
B
R
I
E
F
B
F
F
R
E
A
K
B
F
O
R
B
B
F
O
R
B
+
F
R
E
A
K
B
F
S
I
F
T
B
F
S
U
R
F
B
F
S
I
F
T
F
L
A
N
N
S
U
R
F
F
L
A
N
N
0
1000
2000
3000
4000
5000
6000
Average me per Frame
Average <me per Frame
Time(ms)
Descrip<on:
Abscissa: algorithms (eg. BRISK indicates that the detec<on and extrac<on use the same
algorithm. FAST+BRIEF indicate that detec<on uses the FAST algorithm and the extrac<on uses
the BRIEF algorithm. Besides, BF & FLANN are the feature matching algorithms)
Ordinate: consumed <me
The result is the average of ten pictures and each test picture uses four methods to be tested,
including brightness changes, Gaussian blur, rota<on and scaling.
The consumed <me includes feature detec<on, feature extrac<on and feature matching.
资源评论
