基于51duino智能小车的三维场景重建

Three Dimensional Scene Reconstruction Based on 51duino Smart Vehicle

为了解决光学单目拍摄图像的物体相互遮挡、深度信息缺乏等问题,研究了基于51duino智能小车的三维场景重建方法.智能小车的核心控制器采用开源51duino开发平台,并搭载有单目光学摄像头、超声避障模块、光电寻迹模块与wifi传输模块.小车在行驶过程中,由不同角度拍下被测场景图像,经wifi模块将图像传回用户终端,并在用户终端采用SFM(Structure from Motion)算法得到场景的稀疏点云,再经CMVS(Cluster Multi-View Stereo)/PMVS(Patch-based Multi-View Stereo)稠密重建与泊松表面重建,得到场景的三维图像.通过两组测试验证了所提方法的正确性与可行性,获得了信息更丰富、可读性更强的场景三维图像.

To solve the object occlusion and depth information poverty in monocular optical images,three dimensional scene reconstruction was researched based on 51duino smart vehicle.The open source 51duino platform was adopted for the core controller of smart vehicle,and the peripheral sensors included a monocular optical camera,ultrasonic obstacle model,photoelectric tracing model and wifi transmission model.While the smart vehicle was moving,optical images of the scene were acquired by the monocular camera from different perspectives and transmitted to user terminal via wifi model.At user terminal,SFM algorithm was utilized to obtain 3D sparse point cloud.Then CMVS(Cluster Multi-View Stereo)/PMVS(Patch-based Multi-View Stereo)dense reconstruction and Poisson surface reconstruction were used to obtain 3D images of the scene.At last,indoor and outdoor scenes were tested,demonstrating that the proposed 3D reconstruction based on smart vehicle is correct and feasible.The obtained 3D reconstruction results of the tested scenes are with richer information and greater readability.