Image Stitching Algorithm: An Optimization between Correlation-Based and Feature-Based Method

Image Stitching Algorithm: An Optimization between
Correlation-Based and Feature-Based Method
Moushumi Zaman Bonny
Dept. of Computer Science and Engineering
Jahangirnagar University
Dhaka, Bangladesh
zaman.moushumi@gmail.com
Mohammad Shorif Uddin
Dept. of Computer Science and Engineering
Jahangirnagar University
Dhaka, Bangladesh
shorifuddin@gmail.com
Abstract—Image stitching detects several images of the same
scene and then merges those images to generate a single
panoramic image. This paper presents a framework to compare
different kind of panorama-creation process, such as correlation-
based method and feature-based method with a view to develop
an optimum panorama. The evaluations are done by comparing
the outputs with respect to the original ground truth along with
computation time. We have done simulations by applying these
two approaches to draw a satisfactory resolution.
Keywords-panorama; imagestitching; image features and
feature detection; correlation
I. INTRODUCTION
Image stitching is a mechanism for forming panoramic
images. We know that over-lapping regions of input images
may produce visible seam between images which is caused by
the disparities in camera responses and illuminations of scene
and errors in spatial alignment. Image stitching is a technique
to locate and remove the seams of the overlapping areas of the
input images and then blend together [1].
There are many methods to produce stitched image. Image
stitching is less hardware-intensive but mostly software-based
mechanism to generate a wide field of view image. It captures
many regular pictures to cover up the whole viewing space
and stitches together all the images to build a single image
with a larger field of view [2]. Image alignment or registration
of images, is a prerequisite for image stitching. Alignment or
registration is the process of finding correspondence
considering rotation, translation and scaling between two
images of the same scene.
The categories of image stitching techniques can be
generally classified into intensity based- or direct- and feature-
based-method. Direct method is very functional for stitching
images which have no overlapping region. Feature based
method can be functional in small overlapping region. Image
stitching using correlation is one of the most primitive
stitching techniques, which is basically an intensity-based
approach and suitable for images with overlapping or non-
overlapping region [3], [4].To establish correspondences
among a set of images, feature correspondence between two or
more images are considered necessary, so, it is essential to
identify a set of prominent points in each image [5].
In image registration, it is compulsory to align two or
more images of a scene. The major steps, concerned in image
registration or alignment tasks are: (i) feature detection; (ii)
feature matching based on transformation functions using
corresponding features in images; (iii) reconstruction of
images using derived transformation function. For matching
and recognition, the first step is to detect location of interests
in the images. Then descriptors are computed and these need
to be compared to find the association between images for
matching operation. So, we should use feature detector and a
feature descriptor for extracting features from images [6], [7].
The main objective of this paper is to give a guideline
for optimum stitching to fulfill the satisfaction on diverse
applications. We have structured the rest of the paper as
follows: Section II describes literature review. In Section III,
we have presented a brief depiction of the stitching algorithms.
In Section IV, we have performed the performance evaluation
on the obtained results. We discussed about applications of
image stitching algorithms in Section V. Finally, we have
turned the conclusion of the paper along with some future
research directions in Section VI.
II. LITERATURE REVIEW
Many researchers have already been worked on
image stitching. A thorough review on image stitching is
presented below:
Yang et al [8] proposed an image mosaic method
based on phase-correlation and Harris operator. Wang et al [9]
presented an automatic panoramic image mosaic method
based on graph model.
Adel et al [10] proposed an image stitching system
based on ORB (Oriented FAST (Features from Accelerated
Segment Test) and Rotated BRIEF (Binary Robust
Independent Elementary Features)) technique and
compensation blending. Szeliski [11] proposed a method of
image alignment and stitching. Fatah e t a l [1 2 ] h ad a
p r o p o sa l o f automatic seamless of image stitching
method. Qiuet al [13] presented an algorithm based on SIFT
International Journal of Computer Science and Information Security (IJCSIS),
Vol. 16, No. 6, June 2018
150 https://sites.google.com/site/ijcsis/
ISSN 1947-5500

(Scale Invariant Feature Transform) based feature matching
and transformation parameters. Ostiak et al [14] presented a
fully automated HDR (High Dynamic Range) panorama
stitching algorithm where he used the SIFT based algorithm
for the recognition of the corresponding feature points.
All methods have merits and demerits. With this
view, in the current paper we have performed a comparative
study on feature based- and correlation based-method to
develop an optimum image stitching approach on the basis of
application perspective.
Figure 1. Flow diagram of our methodology.
III. METHODOLOGY
The methodology consists of the following steps:
A. Image Acquisition
The first step is acquisition of images which means
capturing images by camera devices or acquiring from
secondary sources.
B. Alignment
The reason of alignment is to locate are liable alignment
parameter which can decrease the miss-registration between
every pair of images. It is very useful to widen the pair wise
matching criterion [15], [16].
So, here, we need to detect and match features between
images I(n) and I(n-1).Then, we have to estimate geometric
transformation T that maps images 𝐼 to 𝐼 𝑛 as follows[17],
[18],
𝑇 = 𝑇 × 𝑇 × 𝑇 × … 𝑇 𝑛 − × 𝑇 𝑛
In Eq. (1), 𝑇 , 𝑇 , 𝑇 , … , 𝑇 𝑛 − , 𝑇 𝑛 are
the geometric transformations of images
𝐼 , 𝐼 , 𝐼 , … , 𝐼 𝑛 − , 𝐼 𝑛 , respectively.
C. Locating Overlapping Region
Locating the overlapping region is the next step. This is
essential for the consequent features between several images
[20], [21].If the overlapping region exists, then we can have
two types of output: feature-based panorama and correlation-
based panorama. But, if overlap does not exist, our method
stitches the images directly one after another one.
D. Feature-Based Stitching
Generally, for creating panorama, one image i s
s e l e c t e d as the reference and then all the images are
sequentially used for stitching. The result is often known as a
flat panorama [22], [23].
Image stitching can be applied on s e v e r a l projective
layouts, for example, rectilinear projection, where the
panorama is observed lying on a plane (two-dimensional)
crossing a pano-sphere at a point [24]. Rectilinear projection
uses cube faces with cubic mapping for viewing panorama and
it demonstrates a cylindrical projection [25], [26].
After plotting the source pixels against the ultimate
composite surface, it is necessary to blend them for creating a
panorama. Therefore, blending operation is a
prerequisite for image stitching. Feathering, image
pyramid and gradient domain are some familiar blending
procedures [27].Blending operator forms a blend of two input
images of the same size [28],[29]. Feathering is used for
blurring the edges of features. From the two overlapping
images, the average pixel values are evaluated for the blended
regions. Multi-band image blending is another well-known
method that executes in the gradient domain [30],[31],[32].
These magnitudes are used such that the output pixel
values do not surpass the highest pixel value.
The resulting image , is calculated using the
formula,
, = 𝑋 × , + − 𝑋 × ,
In Eq. (2), P1 and P2 are the two input images. In some
applications, X may also be a constant, thus allowing a
constant offset value to be added to a single image. X is the
blending ratio which determines the control of each input
image points into output. X can either be a constant factor for
all pixels in the image or can be determined for each pixel
separately using a mask. The size of the mask must be
identical with the size of the images [33].
E. Correlation-Based Stitching
We do not need to extract features in correlation-based
stitching approach. This process depends on the relation
between two images. It is variant to rotation, scaling and other
Vol. 16, No. 6, June 2018
ISSN 1947-5500

transformation. For accurate preprocessed images, correlation
approach is efficient to apply.
Determining the correlation values is the first step in
this method. Matching interest points in two un-calibrated
images is fundamental problem in computer vision.
Correlations generally used in many applications that require
matching parts of the images [34], [35], [36]. We have
calculated correlation coefficient r between two images
and using the following equation,
𝑟 =
∑ ∑ − ′
− ′
√ ∑ ∑ − ′ ∑ ∑ − ′
In Eq. (3), A′
= mean A andB′
= mean B .
After determining the maximum correlation value of the two
images we have to find the maximum degree of column vector
[37], [38].
Merging is the next step. When the source pixels are
mapped onto the final surface, then these are blended in order
to generate a panorama and the seam line adjusted to reduce
the visibility of seams between two images[39], [40],[41].
F. Comparison
In this step, Feature-based panorama and Correlation-
based panorama are compared with the Ground truth. The
approach will move forward with the better result depending
on the optimization of accuracy.
IV. RESULTS AND DISCUSSIONS
We applied our method on ten-image set and each
image set consists of two images as inputs. We have prepared
two input images (480×475 pixels each, for image set 1 to 8)
from some original images using Microsoft Office Picture
Manager. Besides these, we also used input images of other
sizes (500×475 pixels for image set 9 and 700×475 pixels for
image set 10).All these images have some overlapping regions
as well as repetitions. Then, we have inputted these images to
create the panorama using stitching techniques: feature-based
panorama and correlation-based panorama [39]. All the outputs
are generated using MATLAB R2017a with Microsoft Windows
10 platform, Intel Core i3, 2.00 GHz processor and 4.00GB
RAM. During experimentation, we found that feature-based
method consumes more time but correlation-based method
takes less time. So, if these two methods are compared in term
of speed, the correlation method is faster. Accuracies for the
both methods are almost similar, as the images are mostly
aligned from the initial stage. Figure 2 shows an image (which
is used as a ground truth for the stitched image). This image is
cut into two images with some overlapping and shown in
Figure 3. Figure 4(a) shows the registration process on the
basis of feature points. Figure 4(b) shows the overlapping
region. Figure 4(c) presents the stitched image using
correlation-based method and Figure 4(d) shows the stitched
image for feature-based method. Similar to Figures 2-4, we
have done the same operations in Figures 7-9 for an HDR
image and in Figures 10-12 for a microscopy image. Figure 5
shows two images without any overlap region and Figure 6
presents its corresponding stitched image by direct method
(simple stitching).
TABLE I. PERFORMANCE EVALUATION OF IMAGE STITCHING
TECHNIQUES BASED ON ACCURACY RATE AND TIME
Input
Image and its
size
Correlation-based
Method
Feature-based Method
Accuracy
(%)
Time
(Sec)
Accuracy
(%)
Time
(Sec)
Image Set 1
(480×475)
98.61% 0.3937s 96.44% 0.7166s
Image Set 2
(480×475)
98.86% 0.4672s 96.36% 0.9883s
Image Set 3
(480×475)
99.62% 0.4596s 98.16% 0.9081s
Image Set 4
(480×475)
99.71% 0.4011s 98.82% 0.8092s
Image Set 5
(480×475)
99.79% 0.4222s 99.05% 0.9951s
Image Set 6
(480×475)
99.09% 0.5716s 97.43% 1.6230s
Image Set 7
(480×475)
99.29% 0.8725s 98.53% 1.0128s
Image Set 8
(480×475)
99.29% 0.7938s 97.13% 1.7026s
Image Set 9
(500×475)
98.60% 0.7674s 96.65% 1.1427s
Image Set 10
(700×475)
99.59% 0.7571s 98.52% 0.9608s
We have used the following equation to calculate the
accuracy:
Accuracy = − sum sum | 𝐼 − 𝐼 | /sum sum 𝐼
(4)
In Eq. (4), 𝐼 is the original image, 𝐼 is the resultant image of
the experiment. Table I shows the performance (accuracy and
computation time) of the stitching method.
Figure 2. Original image (treated as ground truth panorama)
Vol. 16, No. 6, June 2018
ISSN 1947-5500

Figure 3. Two input images which are generated from the original image of Fig. 2 with some overlapping.
(a)
(b)
(c)
(d)
Figure 4. Outputs: (a) Aligned/ registred image on the basis of feature points,(b)Overlapping region, (c) Correlation- based approach, (d) Feature-based approach.
Figure 5. Input images
Figure 6. Output images:Direct Method
Vol. 16, No. 6, June 2018
ISSN 1947-5500

Figure 7. Original HDR image (treated as ground truth panorama).
Figure 8. Input images which are generated from the original image of Fig. 7 with some overlapping.
(a)
(b)
(c)
(d)
Figure 9. Outputs: (a) Aligned/ registred image on the basis of feature points,(b)Overlapping region, (c) Correlation- based approach, (d) Feature-based
approach.
Vol. 16, No. 6, June 2018
ISSN 1947-5500

Figure 10. Original microscopyimage of renal cell carcinoma (treated as ground truth panorama).
Figure 11. Input images which are generated from the original image of Fig. 10 with some overlapping.
(a)
(b)
(c)
(d)
Figure 12. Outputs: (a) Aligned/ registred image on the basis of feature points,(b)Overlapping region, (c) Correlation- based approach, (d) Feature-based
approach
Vol. 16, No. 6, June 2018
ISSN 1947-5500

V. APPLICATIONS
Image stitching is mostly used in diverse applications,
such as, image stabilization, HDR (High Dynamic Range)
image mosaicing, high resolution photo mosaics in digital maps
and satellite photos, medical imaging, multiple image super-
resolution, video stitching, object insertion, microscopy image
stitching and group photographs. In this paper, we have
experimented with natural-, HDR- and microscopy-image and
got satisfactory results.
VI. CONCLUSION AND FUTURE WORKS
Image stitching is the process of forming a high
resolution stitched (panorama) image by combining two or
more images together. These input images have some
overlapping region with wide field of view to structure a
segmented panorama. There are different algorithms for
feature detection. Feature-based method uses feature
descriptors and correlation-based method uses correlation
values to create panorama.
In this paper, we have shown a framework to produce
an optimized panorama image on the basis of application
needs. We have thoroughly investigated feature-based and
correlation-based with accuracy and computation time.
Usually, feature-based method needs higher computation time.
But correlation method is suitable if the two images are
aligned that means registered. Features are very useful for
good registration. So, for high quality output in somehow non-
aligned environment feature-based method is the optimum
choice.
High-quality panorama also depends on light
exposure and distortion of lens. So, image calibration can be a
supplementary part between acquisition of images and image
registration. For good visual feeling we have used alpha
blending method that can be replaced by a more robust
technique. Another intricacy is photo distortion that arises
from the perspective transformation and can be improved by
lens distortion correction method.
REFERENCES
[1] L. Juan, G. Oubang, “SURF applied in panorama image stitching,Image
Processing Theory Tools and Applications (IPTA), 2010 2nd
International Conference, pp.495 – 499, Jul 2010.
[2] HyungIlKoo and NamIkCho,“Feature-based Image Registration
Algorithm for Image Stitching Applications on Mobile Devices,” IEEE
Transactions on Consumer Electronics, vol.57, no.3, pp. 1303-1310,
Aug 2011
[3] R. Karthik, A. Annis Fathima,V.Vaidehi,“PanoramicViewCreation using
InvariantMoments and SURF Features,”IEEE International Conference
on Recent Trends in Information Technology (ICRTIT), pp. 376-382,
July 2013.
[4] Deepak Kumar Jain ,Gaurav Sexena, Vineet Kuma rSingh,“Image
mosaicing using corner technique,” IEEE International Conference on
Communication Systems and Network Technologies, pp.79-84, may
2012.
[5] Mathew Brown, David G. Lowe, “Automatic Panoramic Image Stitching
using Invariant Features,” International Journal of Computer Vision, vol.
74, pp.59 – 73, Aug 2007.
[6] BrownM. and Lowe, “Recognising Panoramas,” Ninth IEEE
International Conferenceon Computer Vision, vol. 2,pp.1218-1225,
2003.
[7] Feng Zhao, Qingming Huang, Asn Gao, “Image Matching
ByNormalized Cross-Correlation”, Institute of Computing Technology,
Chinese Academy of Sciences, Beijing, China vol.60, no.1, pp.63-86,
2004.
[8] Fan YANG, Linlin WEI, Zhiwei ZHANG, Hongmei TANG, ”Image
Mosaic Based on Phase Correlation and Harris Operator”, Journal of
Computational Information Systems, vol. 8, pp. 2647–2655, 2012.
[9] Zhicheng Wang and Yufei Chen “An Automatic Panoramic Image
Mosaic method based on Graph Model” In Springer Link. vol. 75, Issue
5, pp. 2725-2740, 2016.
[10] Ebtsam Adel, Mohammed Elmogy, Hazem Elbakry, “Image Stitching
System Based on ORB Feature- Based Technique and Compensation
Blending,” International Journal of Advanced computer Science and
Applications, vol. 6, No. 9, 2015.
[11] Szeliski,“Image Alignment and Stitching,” Tech. rep.,
December10,2006.
[12] Russol Abdel fatah, Dr. Haitham Omer, “Automatic Seamless of
Image Stitching,’’Journal of Computer Engineering and Intelligent
systems,vol. 4, No. 11, 2013.
[13] Pengrui Qiu, Ying Liang and Hui Rong, “Image Mosaics Algorithm
Based on SIFT Feature Point Matching and Transformation Parameters
Automatically Recognizing” 2nd International Conference on Computer
Science and Electronics Engineering , 2013.
[14] Piotr Ostiak, “Implementation of HDR panorama stitching algorithm,”
Semantic Scholar.
[15] Deng.Y. and Zhang.T.,“Generating Panorama Photos”,International
Society for Optics and Photonics, pp. 270-279, 2003.
[16] Xianyong Fang,“Feature Based Stitching of a Clear/Blurred Image
Pair,”IEEE International Conference on Multimedia and Signal
Processing, vol. 1,pp. 146-150, 2011.
[17] Oh-SeolKwonandYeong-HoHa,“Panoramic Video using Scale-Invariant
Feature Transformwith Embedded Color-Invariant Values,”IEEE
Transactions on Consumer Electronics,vol.56,No. 2, May 2010.
[18] K.Shashank, N.Siva Chaitanya, G.Manikanta,Ch.N.V.Balaji, V.
V. S.Murthy,“A Survey and Review Over Image Alignment and
Stitching Methods,” International Journal of Electronics &
Communication technology, vol.5, pp.50-52,march 2014.
[19] A.Levin,A.Zomet,S.Peleg,Y.Weiss,“Seamless image stitching in the
gradient domain,”In Eighth European Conference on Computer Vision
(ECCV2004),(Prague), vol. IV, pp. 377-389, Springer-Verlag, May 2004.
[20] A.Zomet, A.Levin, S.Peleg, Y.Weiss, “Seamless Image Stitching by
minimizing false edges, ”IEEE Transactions on Image Processing, vol.
15, No. 4, pp. 969-977, April 2006.
[21] Feature Based Panoramic Image Stitching, Retrieved from
http://www.mathworks.com/help/vision/examples/feature-based -
panoramic-image-stitching.html
[22] J.Y.Jia,C.K.Tang,“Image Stitching Using Structure Deformation,”IEEE
Transactionson Pattern Analysis and Machine Intelligence, vol. 30, No.
4, pp. 617-631, 2008.
[23] Mclauchlan,P.F.,Jaenicke,A.,Xh,G.G., “Image mosaicing using
sequential bundle adjustment,” Proc. BMVC, pp.751-759, 2000.
[24] ChaoTao, Hanqiu Sun, Changcai Yang, Jinwen Tian, “Efficient Image
Stitching in the Presence of Dynamic Objects and
Structure Misalignment,”Journal of Signal and Information
Processing,vol. 2, pp. 205-210 , 2011.
[25] Rosten.E.and Drummond.T.,“Machine Learning for High-speed Corner
Detection,”EuropeanConferenceonComputerVision,vol.1, pp. 430-
443,2006.
[26] Tao-Cheng Chang,Cheng-AnChien, Jia-Hou Chang, and Jiun-In
Guo,“ALow-Complexity Image Stitching Algorithm Suitable for
Embedded Systems,”IEEE International Conferenceon Consumer
Electronics(ICCE),pp. 197-198, jan 2011.
[27] Patrik Nyman,“Ima g e Stitching using Watersheds and Graph
Cuts,’’ Research Article, Lund University, Sweden.
[28] Ashley Eden,Matthew Uyttendaele, Richard Szeliski, “Seamless
Image Stitching of Scenes with Large Motions and Exposure
Differences,’’ Proceedings of the 2006 IEEE Computer Society
Vol. 16, No. 6, June 2018
ISSN 1947-5500

Conference on Computer Vision and Pattern Recognition, pp. 2498-
2505, June 17-22, 2006.
[29] Panoramic Mosaic Stitching, http://pages.cs.wisc.edu/~lizhang/
courses/cs766-2008f/projects/mosaic/students/bfield/report.html
[30] EbtsamAdel, Mohammed Elmogy, Hazem Elbakry,“ImageStitching
based on Feature Extraction Techniques: A Survey,” International 6th
Journal of Computer Applications, vol.99, no.6, pp.1-8, August 2014.
[31] Ji-Hun Mun, G. Oubong, “Panorama Image Stitching based on ‘SIFT’
Feature Points,” International Conference on Embedded Systems and
Intelligent Technology, pp. 55-56.
[32] Ze-lang Wang, Fang-huaYan, Ya-yuZheng, “An Adaptive Uniform
Distribution SURF for Image Stitching,”IEEE International conference
on Image and Signal Processing (CISP), vol.2, pp.888-892, 2013.
[33] Zhen Hua,Yewei Li,JinjiangLi,“Image Stitch Algorithm Based on SIFT
and MVSC,” IEEE7thInternational Conference on Fuzzy Systems and
Knowledge Discovery,vol. 6, pp. 2628-2632,Aug 2010.
[34] Rosten, E., & Drummond, T., “Machine Learning for High-speed Corner
Detection,” 9th European Conference on Computer Vision, Vol. Part I,
pp. 430-443, Berlin, Heidelberg: Springer-Verlag, 2006.
[35] Pranoti Kale, K.R.Singh, “A Technical Analysis of Image Stitching
Algorithm,” International Journal of Computer Science and Information
Technologies, vol. 6, pp. 284-288, 2015.
[36] A.V.Kulkarni, J.S.Jagtag, V.K.Harpale,“Object recognition with ORB
and its Implementation on FPGA,” International Journal of Advanced
Computer Research, pp. 164-169, 2013.
[37] J. Matas, O. Chum, M. Urban, T. Pajdla,“Robust wide baseline stereo
from maximally stable extremal regions,” Proc. of British Machine
Vision Conference, pp. 384-396, 2002.
[38] Panorama of University from Water Tower Looking
North,http://www.findit.library.yale.edu/?page=3&q=Panorama+image
&search_field=all_fields&utf8=?
[39] ZhengyouZhang,“A Flexible New Technique for Camera
Calibration,”IEEE Transactions on Pattern Analysis and Machine
Intelligence, pp. 1330-1334, Nov 2000.
[40] Preeti Mandle, Bharat Pahadiya,“An Advanced Technique of Image
Matching Using SIFT and SURF,”International Journal of Advanced
Research in Computer and Communication Engineering,Vol. 5, Issue 5,
May 2016.
AUTHORS PROFILE
Moushumi Zaman Bonny received her M.S. and B.Sc. in Computer Science
and Engineering from Jahangirnagar University, Savar, Dhaka. She is now
pursuing her PhD at the same university. Her research interests include
Pattern Recognition, Digital Image Processing and Computer Vision.
Mohammad Shorif Uddin received his Doctor of Engineering degree in
information Science from Kyoto Institute of Technology in 2002, Japan,
Master of Technology Education degree from Shiga University, Japan in 1999,
Bachelor of Electrical and Electronic Engineering degree from Bangladesh
University of Engineering and Technology in 1991 and also MBA in from
Jahangirnagar University in 2013. He began his teaching career as a Lecturer
in 1991 at the Bangladesh Institute of Technology, Chittagong (Renamed as
CUET). He joined in the Department of Computer Science and Engineering of
Jahangirnagar University in 1992 and currently, he is a Professor of this
department. In addition, he is an Adviser of the School of Science and
Engineering, ULAB and Teacher-in-Charge of ICT Cell of Jahangirnagar
University. He served as the Chairman of Computer Science and Engineering
of Jahangirnagar University from June 2014 to June 2017. He undertook
postdoctoral researches at Bioinformatics Institute, Singapore, Toyota
Technological Institute, Japan and Kyoto Institute of Technology, Japan,
Chiba University, Japan, Bonn University, Germany, Institute of Automation,
Chinese Academy of Sciences, China. His research is motivated by
applications in the fields of imaging informatics and computer vision.
Mohammad Uddin is an IEEE Senior Member and also Fellow of Bangladesh
Computer Society and The Institution of Engineers Bangladesh. He has
lectured a good number of undergraduate and graduate courses, he wrote more
than 100 journal and conference papers, three books, technical reports and a
reviewer for many journals and conferences; he organized national and
international conferences and seminars. Three researchers achieved their PhD
degrees under his supervision. He had delivered a remarkable number of
keynotes and invited talks and also wrote many articles for magazines and
newspapers. He is the Editor-in Chief of ULAB Journal of Science and
Engineering, Associate Editor of IEEE Access, General Chair of IJCCI 2018,
IWCI 2016 and TPC Chair/Co-Chair of ICEEICT 2015, IEEE R10 HTC 2017,
EICT 2015 and EICT2017. He holds two patents for his scientific inventions.
He received the Best Paper award in the International Conference on
Informatics, Electronics & Vision (ICIEV2013), Dhaka, Bangladesh and Best
Presenter Award from the International Conference on Computer Vision and
Graphics (ICCVG 2004), Warsaw, Poland. He was the Coach of
Janhangirnagar University ACM ICPC World Finals Teams at USA in 2017
and Morocco in 2015. Besides, he worked as an IT Consultant at Maidas
Financing Limited. He served as Membership development coordinator (2015)
and Conference Coordinator (2016-2017) of IEEE Bangladesh Section.
Currently, He is a Vice Chair (Technical) of IEEE CS Bangladesh Chapter and
a Vice President (Admin) of Bangladesh Computer Society (BCS). He is a
Fellow of IEB and BCS and a Senior Member of IEEE and IACSIT.
Vol. 16, No. 6, June 2018
ISSN 1947-5500

Image Stitching Algorithm: An Optimization between Correlation-Based and Feature-Based Method

Recommended

More Related Content

What's hot (18)

Similar to Image Stitching Algorithm: An Optimization between Correlation-Based and Feature-Based Method (20)

Recently uploaded (20)

Image Stitching Algorithm: An Optimization between Correlation-Based and Feature-Based Method