研究目的
To improve information acquisition efficiency in ocean visualization by proposing a joint imaging method that combines acoustic and optical imaging, addressing the limitations of each method when used independently.
研究成果
The proposed acousto-optical fusion system effectively combines sonar and optical imaging to achieve efficient and detailed ocean visualization. Image enhancement and mosaic processing significantly improve underwater image quality, as evidenced by increased feature point matches and better histogram distributions. Future work should focus on implementing the full system, including underwater acoustic positioning, AUV cluster communication, and network optimization for real-world applications.
研究不足
The experiments are conducted in a controlled pool environment with limited size and depth, which may not fully represent complex ocean conditions. The system relies on acoustic communication, which can be affected by underwater noise and propagation issues. Image processing algorithms may have reduced effectiveness in highly turbid or deep waters.
1：Experimental Design and Method Selection:
The study designs an acousto-optical fusion system with a decision-making layer using sonar for target detection and localization, and an executive layer using AUV-mounted cameras for optical imaging, followed by image enhancement and mosaic processing. Theoretical models include SURF algorithm for feature point extraction and matching, RANSAC for mismatching point screening, and Retinex algorithm for image enhancement.
2：Sample Selection and Data Sources:
A pool experiment is conducted using a sinking ship model as the target in a 4m*3m*1m pool with 0.5m water depth. Images are captured by an underwater camera under natural light during uniform rotation.
3：5m water depth. Images are captured by an underwater camera under natural light during uniform rotation.
List of Experimental Equipment and Materials:
3. List of Experimental Equipment and Materials: Underwater camera (specifications: 800 TVL definition, 92° viewing angle, DC12V 1A power input, working temperature -20° to +50°), pool, wreck model (bow broken block: high 25 cm, wide 25 cm, long 44 cm; stern broken block: high 27 cm, wide 23 cm, long 38 cm).
4：Experimental Procedures and Operational Workflow:
The decision-making layer uses sonar to detect and locate the target, then sends instructions via acoustic communication to an AUV. The AUV-mounted camera captures optical images, which are preprocessed with denoising, color balance, and enhancement (using median filtering and Retinex algorithm), followed by image mosaic using SURF and RANSAC for feature matching and fusion.
5：Data Analysis Methods:
Image quality is assessed by comparing original and enhanced images, counting matched feature points, and analyzing R/G/B histograms for contrast and color distribution.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
