Image stitching algorithm research based on OpenCV

Kaili Chen*, Meiling Wang

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

13 Citations (Scopus)

Abstract

In view of the vast scene and high-resolution image stitching problems, this paper studies a kind of Harris image stitching algorithm which is based on OpenCV configuration environment. First of all, the Harris corner detection is used to extract the feature points. Then Normalized Cross Correlation is used to rough match the feature points and RANSAC algorithm is used to eliminate error matching. Secondly, cylindrical projection transformation model is adopted to realize image registration algorithm. And finally this paper uses an improved weighting average fusion algorithm to fuse images, reducing the computational complexity of image fusion and eliminating seam of stitched images. This algorithm reduces the overlapping rate of images during image stitching and increases the speed of image Registration and Mosaic. It realizes the purpose of giving consideration to both precision and efficiency.

Original languageEnglish
Title of host publicationProceedings of the 33rd Chinese Control Conference, CCC 2014
EditorsShengyuan Xu, Qianchuan Zhao
PublisherIEEE Computer Society
Pages7292-7297
Number of pages6
ISBN (Electronic)9789881563842
DOIs
Publication statusPublished - 11 Sept 2014
EventProceedings of the 33rd Chinese Control Conference, CCC 2014 - Nanjing, China
Duration: 28 Jul 201430 Jul 2014

Publication series

NameProceedings of the 33rd Chinese Control Conference, CCC 2014
ISSN (Print)1934-1768
ISSN (Electronic)2161-2927

Conference

ConferenceProceedings of the 33rd Chinese Control Conference, CCC 2014
Country/TerritoryChina
CityNanjing
Period28/07/1430/07/14

Keywords

  • Harris Corner Detection
  • NCC
  • OpenCV
  • Weighted Fusion

Fingerprint

Dive into the research topics of 'Image stitching algorithm research based on OpenCV'. Together they form a unique fingerprint.

Cite this