2025 Vol.8 Aug.N04

• Title: Research on Optimization of Image Compression Algorithm Based on Deep Learning
• Name: Lei Liu
• Company: Chengdu College of University of Electronic Scienceand Technology of China，Chengdu  611731,China
• Abstract:

  This article proposes a deep learning based image compression algorithm optimization method to address the potential shortcomings of current image compression algorithms in processing images with special textures, structures, or noise characteristics. We adopt a generative adversarial network model to automatically learn the potential distribution and feature representation of images through adversarial learning mechanisms, thereby more effectively encoding and decoding images. At the same time, based on traditional encoders for encoding and decoding, and adding total variation constraints to the output of GAN to improve compression performance. This total variation constraint can suppress noise in the image while preserving the edge structure characteristics of the original image. The experimental results show that compared to other methods, the image compression effect of our method is superior, with excellent performance in compression ratio, peak signal-to-noise ratio, FSIM and structural similarity, providing an efficient and high-quality solution for the field of image compression.

  
  

• Keyword: Deep learning; Image compression; Generate adversarial networks; Algorithm optimization; Image quality
• DOI: 10.12250/jpciams2025090804
• Citation form: Lei Liu.Research on Optimization of Image Compression Algorithm Based on Deep Learning[J]. Computer Informatization and Mechanical System,2025,Vol.8,pp.18-21

[1]Liu Qiang, Wu Depei. Simulation of Lossless Anti-Distortion Compression Algorithm for High Frame Rate Virtual Image[J]. Computer Simulation,2023,40(12):252-255,352.

codecs[J].Signal Processing Image Communication, 2020, 82:1157741.1-1157741.10.

[2]Mishra D , Kumar A , Rathor V S ,et al.Hybrid technique for crop image compression using discrete wavelet transform and sparse singular vector reconstruction[J].Computers and electronics in agriculture, 2023, 215(000):108391.1-108391.12.

[3]Ouahioune M , Ameur S , Lahdir M .Enhancing hyperspectral image compression using learning-based super-resolution technique[J].Earth Science Informatics, 2021(6):1-11.

[4]Hajihashemi V , Najafabadi H E , Gharahbagh A A ,et al.A novel high-efficiency holography image compression method, based on HEVC, Wavelet, and nearest-neighbor interpolation[J].Multimedia Tools and Applications, 2021:1-14.

[5]Alkishriwo O A S .Image compression using adaptive multiresolution image decomposition algorithm[J].IET Image Processing, 2020, 14(6):3572-3578.

[6]Karaca A C , Kara O ,Mehmet Kemal Güllü.MultiTempGAN: Multitemporal multispectral image compression framework using generative adversarial networks[J].Journal of Visual Communication and Image Representation, 2021, 81:103385.1-103385.12.

[7]Liu, DeyangHuang, XinpengZhan, WenfaAi, LiefuZheng, XinCheng, Shulin.View synthesis-based light field image compression using a generative adversarial network[J].Information Sciences: An International Journal, 2021, 545(1):1-14.

[8]Wang X , Zhai Y , Ma X ,et al.Low-light image enhancement based on GAN with attention mechanism and color Constancy[J].Multimedia Tools & Applications, 2024, 83(1):3133-3151.