An Object Reconstruction Method Based on Point Cloud with Range Camera

We present an object reconstruction approach based on point cloud. To accurately estimate sensor pose, a combination of frame-to-frame registration and frame-to-model registration is used in our approach with integration to produce the model of reconstructed object. The global optimization used in the approach distributes residual registration errors through the sensor trajectory. Our results demonstrate that the approach is fast and robust to reconstruct objects with a consumer-grade range camera.