Deep Learning Image Compression Method Based On Efficient Channel-Time Attention Module.

Area of Science:

• Electrical Engineering
• Computer Vision
• Data Compression

Background:

• Remote monitoring of transmission lines is crucial for power system stability.
• Weak network signals pose challenges for data transmission and storage in monitoring applications.
• Traditional image compression methods struggle with high-resolution imagery quality and efficiency.

Purpose of the Study:

• To develop an advanced deep learning-based image compression method.
• To improve compression efficiency and reconstruction quality for remote monitoring data.
• To address limitations of existing methods in constrained network environments.

Main Methods:

• Proposed a novel Efficient Channel-Temporal Attention Module (ETAM).
• Integrated Efficient Channel Attention (ECA-Net) and Temporal Attention Module (TAM) within ETAM.
• Employed deep learning for joint spatial and temporal feature extraction.

Main Results:

• ETAM significantly outperformed traditional and state-of-the-art deep learning compression techniques.
• Achieved superior performance across PSNR, SSIM, and LPIPS metrics.
• Demonstrated excellent preservation of fine-grained details and textures at high compression ratios on the STN PLAD dataset.

Conclusions:

• The ETAM method offers a practical solution for efficient, high-quality image compression.
• ETAM shows significant potential for applications like transmission line monitoring under limited network conditions.
• The approach effectively enhances image reconstruction quality and data handling efficiency.