On Optimal Multi-Sensor Network Configuration for 3D Registration

Multi-sensor networks provide complementary information for various tasks like object detection, movement analysis and tracking. One of the important ingredients for efficient multi-sensor network actualization is the optimal configuration of sensors. In this work, we consider the problem of optimal configuration of a network of coupled camera-inertial sensors for 3D data registration and reconstruction to determine human movement analysis. For this purpose, we utilize a genetic algorithm (GA) based optimization which involves geometric visibility constraints. Our approach obtains optimal configuration maximizing visibility in smart sensor networks, and we provide a systematic study using edge visibility criteria, a GA for optimal placement, and extension from 2D to 3D. Experimental results on both simulated data and real camera-inertial fused data indicate we obtain promising results. The method is scalable and can also be applied to other smart network of sensors. We provide an application in distributed coupled video-inertial sensor based 3D reconstruction for human movement analysis in real time.