以目标为中心的MCRBR视点空间投影方法

OBJECT-CENTRED VIEW SPACE PROJECTION METHOD WITH MULTI-CAMERA REVOLVED BATCH RENDERING

三维目标的姿态随视点变化而不同,选择适当角度范围的多尺度投影图像为代表,建立三维目标的完备特征库,能够提高基于图像的三维目标识别率。以目标为观察中心,基于3ds MAX提出一种多摄像机旋转批量渲染的三维目标视点空间投影方法。首先建立目标的三维模型,设置目标模型位置和顶视图摄像机位置,然后利用max层级命令面板锁定其中一个轴向的角度,以等采样间隔复制摄像机,设置半球面运动路径,并约束到相应摄像机,最后设置投影图像的输出参数并建立批处理脚本,实现三维目标视点空间任意投影图像的自动保存。

The attitudes of three dimensional objects vary with the changes of view. Choosing the images to be multi-scale projected in appropriate angles range as the representation to establisb a complete features base for tbree dimensional objects, which can improve the recognition rate of image-based 3D objects. Taking object as the observation centre, and based on 3ds MAX platform, we offer a view space projection method for three dimensional objects with multi-camera revolved batch rendering. First, we construct three dimensional models for the ob- jects, set up the positions for model and top view camera. Then using max level to order the panel sticking to one of the axis angle, and copy the cameras with equal interval. We also set up the moving path of hemisphere, and constraint it to corresponding camera. In the end, we set up output parameters of the projected images as well as establish batch-processing script, and achieve the automatic preservation of random projection images of three dimensional objects view space.