Relative Pose Based Redundancy Removal: Collaborative RGB-D Data Transmission in Mobile Visual Sensor Networks

Area of Science:

• Robotics and Sensor Networks
• Computer Vision and 3D Mapping
• Wireless Communication Efficiency

Background:

• Mobile RGB-D sensor networks are crucial for 3D mapping in various applications.
• Bandwidth constraints in wireless networks limit data transmission efficiency and increase power consumption.
• Existing methods often transmit redundant visual and depth information, leading to inefficiencies.

Purpose of the Study:

• To introduce a novel Relative Pose based Redundancy Removal (RPRR) scheme.
• To enhance wireless channel usage efficiency and reduce power consumption in mobile RGB-D sensor networks.
• To enable rapid 3D mapping in bandwidth-constrained scenarios, particularly for static environments.

Main Methods:

• The RPRR scheme operates in a multiview scenario with multiple RGB-D sensors.
• It identifies and removes redundant visual and depth information by analyzing relative sensor poses.
• The processed, non-redundant data is then transmitted, conserving bandwidth and power.

Main Results:

• Experimental results demonstrate a 250% improvement in wireless channel utilization.
• Battery consumption is reduced by 50% compared to independent sensor image transmission.
• The RPRR scheme effectively minimizes data redundancy in multiview 3D mapping.

Conclusions:

• The Relative Pose based Redundancy Removal (RPRR) scheme significantly improves the efficiency of mobile RGB-D sensor networks.
• RPRR is highly suitable for applications requiring rapid 3D mapping in static, bandwidth-constrained environments, such as disaster recovery.
• The scheme offers substantial benefits in terms of wireless channel usage and power savings.