Research on fast Fourier transforms algorithm of huge remote sensing image technology with GPU and partitioning...

Area of Science:

• Computer Science
• Remote Sensing
• Image Processing

Background:

• Fast Fourier Transforms (FFT) are fundamental to remote sensing image processing.
• Increasing data volumes from hyperspectral, high-resolution, and high-temporal-resolution sensors necessitate efficient processing techniques.
• Existing FFT libraries like CUFFT and FFTW face memory limitations with large datasets.

Purpose of the Study:

• To propose a novel algorithm for efficiently processing huge remote sensing images using FFT.
• To overcome the out-of-memory and memory overflow issues encountered with conventional FFT methods.
• To enhance the effectiveness and speed of remote sensing image processing tasks.

Main Methods:

• Development of a CPU-based Huge Remote Fast Fourier Transform (HRFFT) algorithm incorporating partitioning technology.
• Improvement of the FFT algorithm within the CUFFT function library.
• Experimental validation using CCD images from the HJ-1A satellite.

Main Results:

• The HRFFT algorithm successfully resolves out-of-memory and memory overflow problems for large remote sensing images.
• Experimental results demonstrate the algorithm's rationality and effectiveness.
• Significant improvements in image processing effects and computational speed were achieved.

Conclusions:

• The proposed HRFFT algorithm offers a viable solution for processing massive remote sensing datasets.
• This method enhances the efficiency and accuracy of remote sensing image analysis.
• The HRFFT algorithm contributes to advancements in handling large-scale remote sensing data processing.