Similarity score of two images using different measures.

Appana, Varshini; Guttikonda, Tulasi Manasa; Shree, Divya; Bano, Shahana; Kurra, Himasri

Abstract

In the field of computer vision and image processing, image similarity has been a central concern for decades. If you compare two pictures, Image Similarity returns a value that tells you how physically they are close. A quantitative measure of the degree of correspondence between the images concerned is given by this test. The score of the similarity between images varies from 0 to 1. In this paper, ORB (Oriented Fast Rotated Brief) algorithm is used to measure the similarity and other types of similarity measures like Structural Similarity Index (SSIM), pixel similarity, Earth mover's Distance are used to obtain the score. When two images are compared, it shows how much identical (common) objects are there in the two images. So, the accuracy or similarity score is about 87 percent when the two images are compared.

Citation

APPANA, V., GUTTIKONDA, T.M., SHREE, D., BANO, S. and KURRA, H. 2021. Similarity score of two images using different measures. In Proceedings of the 6th International conference on inventive computation technologies (ICICT 2021), 20-22 January 2021, Coimbatore, India. Piscataway: IEEE [online], pages 741-746. Available from: https://doi.org/10.1109/ICICT50816.2021.9358789

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 6th International conference on inventive computation technologies (ICICT 2021)
• Start Date: Jan 20, 2021
• End Date: Jan 22, 2021
• Acceptance Date: Jan 20, 2021
• Online Publication Date: Feb 26, 2021
• Publication Date: Dec 31, 2021
• Deposit Date: Sep 20, 2023
• Publicly Available Date: Sep 20, 2023
• Publisher: Institute of Electrical and Electronics Engineers (IEEE)
• Peer Reviewed: Peer Reviewed
• Pages: 741-746
• ISBN: 9781728185026
• DOI: https://doi.org/10.1109/ICICT50816.2021.9358789
• Keywords: Image recognition; Image processing; Machine learning; Pixel similarity
• Public URL: https://rgu-repository.worktribe.com/output/2064012