Evaluation of Local Feature Detectors for the Comparison of Thermal and Visual Low Altitude Aerial Images

Divya Lakshmi Krishnan; Rajappa Muthaiah; Anand Madhukar Tapas; Krithivasan Kannan

doi:10.14429/dsj.68.11233

Divya Lakshmi Krishnan Sastra Deemed to be University, Thanjavur 613401
Rajappa Muthaiah Sastra Deemed to be University, Thanjavur 613401
Anand Madhukar Tapas DRDO Defence Research and Development Laboratory, Hyderabad 500 058
Krithivasan Kannan Sastra Deemed to be University Thanjavur 613401

DOI: https://doi.org/10.14429/dsj.68.11233

Keywords: Local features, Aerial images, Thermal images, Image matching, Affine invariant, Image registration

Abstract

Local features are key regions of an image suitable for applications such as image matching, and fusion. Detection of targets under varying atmospheric conditions, via aerial images is a typical defence application where multi spectral correlation is essential. Focuses on local features for the comparison of thermal and visual aerial images in this study. The state of the art differential and intensity comparison based features are evaluated over the dataset. An improved affine invariant feature is proposed with a new saliency measure. The performances of the existing and the proposed features are measured with a ground truth transformation estimated for each of the image pairs. Among the state of the art local features, Speeded Up Robust Feature exhibited the highest average repeatability of 57 per cent. The proposed detector produces features with average repeatability of 64 per cent. Future works include design of techniques for retrieval of corresponding regions.

Author Biographies

Divya Lakshmi Krishnan, Sastra Deemed to be University, Thanjavur 613401

Ms Divya Lakshmi K. is currently doing her full time research in Image Registration at School of Computing SASTRA Deemed to be University, Thanjavur. She received B.E (Electronics and Communication) and ME (Computer Science Engineering with specialisation in Knowledge Engineering and Computational Linguistics). Her interests include programming, image processing and machine learning.

In the current study, she contributed to the design of the feature detectors, their software implementation and testing.

Rajappa Muthaiah, Sastra Deemed to be University, Thanjavur 613401

Mr R. Muthaiah is working as Professor in SASTRA Deemed to be University, Thanjavur, India. He has around 50 International Journal papers and 5 International Conference papers. He is a member of IE and AECE. His research interest includes image processing, VLSI and Network Security.

In the current study, he contributed to the literature survey, data collection.

Anand Madhukar Tapas, DRDO Defence Research and Development Laboratory, Hyderabad 500 058

Mr Anand M. Tapas is working as a Scientist at the Defence Research Development Laboratory, Hyderabad. Earlier, he had worked in Research Centre Imarat, Hyderabad and Automation Division of TISCO and Radio Astronomy Group of TIFR. He is involved in system engineering and simulations.

He provided the initial idea for the study and helped to shape the research, analysis and manuscript.

Krithivasan Kannan, Sastra Deemed to be University Thanjavur 613401

Mr Kannan K. is working as Professor (Department of Mathematics), Dean, School of Humanities and Sciences, in SASTRA University, Thanjavur, India. He is a member of the Ramanujam Mathematical Society. His research interest includes image processing, hypergraphs.