High-density single shot 3D sensing using adaptable speckle projection system with varying preprocessing

摘要

The speckle pattern is highly prevalent in stereo-photogrammetry systems. The speckles generated by the laser and the diffuser are influenced by the subjective noise and change the appearance with the distance. In addition to this, the density of the 3D point cloud is a major stumbling block when only one pattern for reconstruction is allowed. In this work, we propose a system to counter the aforementioned problems. When the distance of the object is changed, the proposed system adapts itself to change the speckle size back to the required optimal size for the algorithm. The speckle size is varied by changing the position of the diffuser using a motorized linear stage. We propose a preprocessing method to reduce the subjective speckle-noise based on histogram equalization and edge preserving contrast enhancement. The amount of preprocessing varies with the distance of the object for the optimization of the 3D reconstructions. Experiments are performed to realize the effect of new preprocessing and variable preprocessing methods. The results indicate that the density of the 3D point is better than the Kinect and previous single shot method and comparable to the multiple shot professional 3D scanner.