GPU Accelerated Image Processing Technology: Architectural Features, Parallel Optimization and Deep Learning Applications

DOI

10.2991/978-94-6463-821-9_40How to use a DOI?

Keywords

GPU; Image Processing; CUDA; CNN

Abstract

This paper systematically discusses the technical basis and optimization methods of GPU-accelerated image processing, focusing on the analysis of GPU architectural features and CUDA and OpenCL parallel computing platforms. It also covers GPU optimization methods in image defogging and enhancement, frequency domain transform (FFT, DCT), feature detection (SIFT) and deep learning driven image processing. The experimental data shows that the parallel optimization strategy based on GPU, in which the computational efficiency of the defogging algorithm is improved by 10 times, the performance of FFT reaches 1000GFlops, and the acceleration ratio of SIFT is up to 121.99 times. Further combining with deep learning scenarios, GPU-based CNN training optimization improves the GEMM operation throughput by 1.97 times. Furthermore, it constructs the OUR-GAN framework to achieve 16k image generation while reducing memory usage to 12.5GB. The study proves that GPU-based software and hardware co-optimization can significantly improve the efficiency of image processing and provide technical support for real-time applications and large-scale computation.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Shaowei Chen
PY  - 2025
DA  - 2025/08/31
TI  - GPU Accelerated Image Processing Technology: Architectural Features, Parallel Optimization and Deep Learning Applications
BT  - Proceedings of the 2025 2nd International Conference on Mechanics, Electronics Engineering and Automation (ICMEEA 2025)
PB  - Atlantis Press
SP  - 386
EP  - 393
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6463-821-9_40
DO  - 10.2991/978-94-6463-821-9_40
ID  - Chen2025
ER  -