PlumX Metrics
Embed PlumX Metrics

An Automatic Algorithm for Stitching Multi-Sequence Retinal Images

SSRN, ISSN: 1556-5068
2022
  • 0
    Citations
  • 394
    Usage
  • 0
    Captures
  • 0
    Mentions
  • 0
    Social Media
Metric Options:   Counts1 Year3 Year

Metrics Details

  • Usage
    394
    • Abstract Views
      328
    • Downloads
      66
  • Ratings
    • Download Rank
      696,744

Article Description

Background and Objective: Registration of retinal color fundus images which were taken at different times, from different perspectives is a critical prerequisite for diagnosis of retinopathy or other eye diseases. For the sake of automatic stitching of multiple retinal images, which can improve the speed of image registration while ensuring the accuracy of matching, a novel automatic registration algorithm of stitching multi-sequence retinal image is proposed in the essay. Methods: The method based on SURT algorithm segments retinal images into small regions, where the area is independent of each other, and then extracts saliency features. The geometric relationship between the sequence retinal image and the reference sequence image is then estimated from the offset which was been calculated from the matched feature point pairs via MODE algorithm. Finally, we use the maximum value fusion to obtain precise fusion image seamlessly. The main innovation of this algorithm is that it is an available avenue for stitching retinal color fundus from multimodality automatically with high speed and accuracy, without the need to adjust parameters. Results: Some experiments demonstrate that the method is highly accurate image stitching, with significantly solved the errors caused by the limitation of the splicing sequence and the accumulation of splicing errors. And the results indicate that the novel algorithm outperforms on private and public medical image database.

Bibliographic Details

Han Yang; Baicheng Li; Chunbo Wu; Lingling Chen; Yuan Liu; Zhensheng Gu

Elsevier BV

Multidisciplinary; multi-sequence retinal image; Image fusion; image registration; SURF feature

Provide Feedback

Have ideas for a new metric? Would you like to see something else here?Let us know