相对于普通灰度和彩色二维图像, 深度图像可以得到物体的三维信息, 使视觉识别和人机交互更加智能。国内外目前还没有低成本、公开的实时获取高质量深度图的方法。本文在对散斑图获取深度图原理研究的基础上, 采取激光散斑的方式, 运用块匹配的方法给出了一种大范围深度图的获取方法。首先, 从原理上验证了块匹配方法的可行性; 然后, 分别从理论和实验两个方面对深度图的计算公式进行了推导和验证; 再次, 对深度图恢复过程进行了详细叙述, 包括散斑图像的预处理和块匹配的过程; 最后, 给出了运用该块匹配方法得到的实验数据。实验结果表明, 本文方法在物体距离相机50 cm左右时精度可以达到5 mm, 200 cm时精度可以达到5 cm, 可以满足室内大部分对象的识别要求。

Compared with the ordinary 2D gray and color images, 3D information of target can be obtained by depth image, making the visual identification and human-computer interaction more intelligent. There is no open and real-time method to obtain high quality depth image at low cost at home and abroad. In this paper, based on the principle research on obtaining the depth image by laser speckle, a way to obtain wide range of depth image is presented by laser speckle and the method of template matching. Firstly, the feasibility of the method of template matching is proved in this paper. Then, the formula of depth image is calculated and verified from both theory and experiments. After that, the process of recovering depth image is described in detail, including pretreatment of speckle image and the process of block matching. Finally, the experimental data obtained by this template matching method are given. The experimental results show that the precision can reach 5 mm when the objects is 50 cm far from the camera and 5 cm when the object is 200 cm far from the camera by using the method proposed in this paper, which can satisfy the reguirements of most indoor objects' recognition.