SharpGAN: Dynamic Scene Deblurring Method for Smart Ship Based on Receptive Field Block and Generative Adversarial...

Area of Science:

• Computer Vision
• Artificial Intelligence
• Marine Technology

Background:

• Complex marine environments pose challenges for vision-based object detection on smart ships.
• Motion blur in images degrades the performance of navigation systems, impacting safety.

Purpose of the Study:

• To develop an effective image deblurring method for smart ship navigation.
• To enhance the accuracy and reliability of object detection algorithms in adverse marine conditions.

Main Methods:

• Proposed SharpGAN, a generative adversarial network (GAN) based image deblurring method.
• Integrated receptive field block net (RFBNet) for enhanced feature extraction.
• Introduced a novel feature loss combining multi-level image features.
• Utilized a lightweight RFB-s module for improved real-time performance.

Main Results:

• SharpGAN achieved superior deblurring performance compared to existing methods.
• The method demonstrated improvements in both subjective visual quality and objective evaluation metrics.
• Enhanced feature similarity between restored and sharp images was observed.
• High correlation with physical model-based deblurring methods was confirmed.

Conclusions:

• SharpGAN offers an efficient and effective solution for deblurring images in smart ship navigation.
• The proposed method significantly improves the robustness of vision sensors in complex marine environments.
• This technology contributes to safer and more reliable autonomous maritime operations.