Multimodal fusion image enhancement technique and CFEC-YOLOv7 for underwater target detection algorithm research

Area of Science:

• Marine technology
• Computer vision
• Image processing

Background:

• Underwater environments present significant challenges for image clarity due to complex conditions.
• Severe static and dynamic blurring in underwater images hinders accurate target recognition and detection.

Purpose of the Study:

• To develop a robust method for enhancing underwater images to improve target recognition accuracy.
• To address both static and dynamic blurring issues in underwater imagery.

Main Methods:

• Adaptive color compensation and improved Multi-Scale Retinex (MSR) for static blur.
• Restormer network for dynamic blur removal (camera shake, out-of-focus, motion).
• Proposed CFEC-YOLOv7 network with CCBC module, ECLOU loss, and FasterNet for target recognition.

Main Results:

• The proposed underwater image enhancement method significantly reduces blurring.
• Qualitative and quantitative analyses confirm the effectiveness of the enhancement techniques.
• The CFEC-YOLOv7 model demonstrates excellent performance in underwater target recognition tasks.

Conclusions:

• The developed underwater image enhancement and target recognition methods are effective for complex underwater environments.
• The CFEC-YOLOv7 model shows strong adaptability and potential for practical applications in underwater detection.