A Computational Intelligence Perspective on Multimodal Image Registration for Unmanned Aerial Vehicles (UAVs)
Remote Sensing (RS) applications generally require robustness, stability, accuracy, promptness, and a high autonomy level to simplify the Big Data (BD) processing in real-time. Image Registration (ImR) is among the most employed RS tasks. ImR transforms d
- PDF / 769,588 Bytes
- 24 Pages / 439.37 x 666.142 pts Page_size
- 75 Downloads / 171 Views
Abstract Remote Sensing (RS) applications generally require robustness, stability, accuracy, promptness, and a high autonomy level to simplify the Big Data (BD) processing in real-time. Image Registration (ImR) is among the most employed RS tasks. ImR transforms different groups of images into a coordinate system that allows overlaying two or more images from the same scene acquired with various sensors and/or taken at different times and angles. The original imageries must be normalized and geometrically aligned to create an ample image containing information from all the separate images. ImR is a crucial step when one has several views and a myriad of sensors that must be fused. The BD aspect of Multimodal Image Registration (MIR) is related to the idea of multispectral and hyperspectral imaging, which involve a vast amount of frequency bands. BD from different sources assist the decision-making processes and create additional more massive datasets for the longterm tracking of various phenomena. This chapter focuses on the MIR from infrared and optical sensors relying on the Particle Swarm Optimization (PSO) class of algorithms. These computational intelligence procedures circumvent problems related V. V. Estrela (B) · M. A. de Jesus Department of Telecommunications, Fluminense Federal University (UFF), Duque de Caxias, RJ CEP 25086-132, Brazil e-mail: [email protected] M. A. de Jesus e-mail: [email protected] N. Razmjooy Department of Electrical Engineering, University of Tafresh, Tafresh, Iran e-mail: [email protected] A. C. B. Monteiro · R. P. França · Y. Iano Nuclear Instrumentation Laboratory, Federal University of Rio de Janeiro, Centro de Tecnologia (CT), Bloco I, sala I-133. Ilha do Fundão, Rio de Janeiro, RJ CEP 21941-972, Brazil e-mail: [email protected] R. P. França e-mail: [email protected] Y. Iano e-mail: [email protected] © Springer Nature Switzerland AG 2021 N. Razmjooy et al. (eds.), Metaheuristics and Optimization in Computer and Electrical Engineering, Lecture Notes in Electrical Engineering 696, https://doi.org/10.1007/978-3-030-56689-0_13
251
252
V. V. Estrela et al.
to multiresolution methods and the high computational cost of hard optimization methods. Keywords Multimodal image registration · Remote sensing · Intelligent agents · Hybrid algorithms · Unmanned aerial vehicle · Particle swarm optimization · Surveillance · Image fusion
1 Introduction Ideally, Remote Sensing (RS) applications require accuracy, stability, robustness, speed, and a high level of autonomy to expedite the processing of enormous volumes of data in real-time [1–4]. Image Registration (ImR) is the procedure of transforming different image sets into a suitable and more general coordinate system. This strategy permits to overlay two or more images from various probing equipment or sensors at different times and angles, or the same scene to geometrically normalize and align them for analysis is a crucial step where information from multiple images must be combined
Data Loading...
