Methodology for data fusion of multichannel images

Abstract

Relevance of the research topic. In modern research and practical application of multi-channel images obtained from various sources, such as aerospace sensors and medical imag-ing systems, there is a need for effective data fusion. In the aerospace industry, combining images with different spectral characteristics provides more accurate information about Earth's surface. In the medical field, integrating images of different modalities helps improve diagnostic accuracy. However, these processes are complicated by differences in spatial resolution, scale, spectral ranges, and noise characteristics of the data.
Problem statement. It is necessary to develop a method for combining images with dif-ferent spectral characteristics to obtain more accurate information about the Earth's surface and medical objects.
Research goal and methods. The goal is to develop and improve a method for combining multi-channel images to improve the accuracy of analysis, increase information content and reduce the impact of noise. The proposed method is based on the use of mathematical models for harmonizing heterogeneous data, including algorithms for preliminary preprocessing, spectral analysis and machine learning. The adaptation of approaches for specific tasks is considered: combining remote sensing data of the Earth for cartographic analysis and merging medical images of different modalities (for example, MRI, CT and ultrasound) for diagnostic purposes.
Results and key conclusions. The developed methodology is universal and can be applied in many areas, including environmental monitoring, urban planning, medical diagnostics, and robotics. The proposed methodology is based on the use of mathematical models to harmonize heterogeneous data and includes preliminary preprocessing, spectral analysis and machine learning. The approaches are adapted for combining remote sensing data of the Earth and fusing medical images for diagnostic purposes. The experiments conducted demonstrate an increase in the quality of the results when using the proposed approach. The results obtained confirm that the integration of multi-channel images allows us to achieve better detail, reduce object identification errors and provide more accurate visualization. The developed methodology is universal and can be applied in many areas, including environmental monitoring, urban planning, medical diagnostics and robotics.