SLIDE 1
2D map construction of images from the Newest Generation of - - PowerPoint PPT Presentation
2D map construction of images from the Newest Generation of - - PowerPoint PPT Presentation
2D map construction of images from the Newest Generation of Endoscopy System Rahman Attar 11/3/2015 Wireless Capsule Endoscopy (WCE) size of a large vitamin A: PillCam SB 3 (Israel); B: MiroCam (South Korea); C: EndoCapsule (USA); D:
SLIDE 2
SLIDE 3
Some characteristics of WCE
- It includes two small colour camera wirelessly transmit thousands of
images during its journey through Gastrointestinal tract.
- An average 8-hour WCE experiment consists of approx. 50000 frames !
- Time required by experienced gastroenterologists to read it can range
from 45 minutes to several hours !
- The quality of WCE images is low.
Capsule Endoscopy
SLIDE 4
Newest Generation: Micro-Ball Endoscopy System six cameras are set in the Micro-Ball
SLIDE 5
Methodology
SLIDE 6
Wireless endoscopic image analysis GI tract Lena
SLIDE 7
Image Registration Image Registration is a vital process which determines the most precise match between two images of the same scene. Image Registration methods can be classified into: 1- intensity based methods 2- feature based methods
SLIDE 8
Algorithm analysis and design The main problem to be solved is formulated in the following equation, in order to complete image registration.
I1 I2 I3 I4 In ... I1 I2 I3 I4 In ...
1 2
{ , ,..., }
n
I I I Ι
result1 result2 Input Image ¡preprocessing, ¡ judgement,local ¡registration Global ¡registration
)) )) ' , ' ( ( ( , )) , ( ( ( arg ~
, , , ,
,
- so
ta j i
CA GI
- v
so j so j i CA GI
- v
ta i ta T j i
d d I f T d d I f S
- pt
T
Ω Ω
=
{ }
)) ( ~ , ,..., , (
2 1
p T I I I I GR I
r n map =
) ( arg ) ( ~
r r
T
- ptD
p T =
2 , 1
~ T
3 , 1
~ T
4 , 2
~ T
n
T ,
2
~
n
T ,
3
~
I1 I2 I3 I4 In ...
2 , 1
~ T
3 , 1
~ T
4 , 2
~ T
n
T ,
3
~
I1 I2 I3 I4 In ...
2 , 1
~ T
3 , 1
~ T
4 , 2
~ T
n
T ,
2
~
...
SLIDE 9
Proposed 2D representation method Image Preprocessing Colour space, Image enhancement, etc. Image Rectification Based on rotation and using Attitude Sensing Unit Image Registration Based on SIFT and MI
SLIDE 10
Image Preprocessing
- A. Attar, Xiang Xie, Chun Zhang, Zhihua Wang, and Shigang Yue,
"Wireless Micro-Ball Endoscopic Image Enhancement Using Histogram Information", 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC’14), Chicago, USA, August 2014.
SLIDE 11
Image Registration This scheme consists of : 1- preregistration process scale-invariant feature transform (SIFT) 2- fine-tuning process maximization of mutual information (MMI)
SLIDE 12
Image Registration Correspondence keypoints are extracted and matched together
SLIDE 13
Image Registration Applying the proposed method on 28 consecutive images
SLIDE 14