中文摘要：

针对工业产品质量检测过程中产品三维表面的重建问题，提出一种基于多目立体视觉三维重建方法．设计了一套由八个直线分布的工业相机构成的三维重建系统方案．首先通过图像采集模块，在八个不同方向对目标物体进行图像采集．其次对采集到的图像进行预处理，其中包括图像背景抑制和目标物体分割．然后通过相机标定模块，对八个相机进行标定，获得它们的内外参数，并结合Harris角点检测及高斯差分检测算法对预处理后的图像实现特征点提取．在此结果上，再利用三角形法对提取到的特征点进行匹配和校正．最后采用泊松表面重建方法准确地获取和优化角点，并找到角点特征的匹配点，从而对物体进行三维表面的精确重建．实验结果表明，设计的系统能够重建出静止物体的局部三维表面，重建结果中的物体表面完整，结构清晰，表面上的字符重建完整，能够很好地进行识别．

英文摘要：

To solve the problem of three-dimensional reconstruction of the surface of the product in the process of industrial products quality inspection, a method based on multi-view stereo vision was proposed. A system of three-dimensional reconstruction program with eight linear industrial cameras was designed. Firstly, through the image acquisition module, the images of the target object were acquired in eight different directions. Secondly, those images were used to preprocess, including image background suppression and target object segmentation. Then the internal and external parameters of eight cameras were got by the camera calibration module, and combined with Harris corner detection and Difference of Gaussian detection algorithm, feature point extraction was achieved for the pretreatment pictures. Based on these results, the extractive the feature points were matched and checked by the triangular method. Finally, after accurately obtaining and optimizing the angle points, finding a corner feature matching points, the surface of object was reconstructed accurately by the Poisson surface reconstruction method. The experimental results show that this system can reconstruct the three-dimensional surface of the static object, the surface of the object is integrity and its structure is clear in the reconstruction results. The characters on the surface are reconstructed completely, which can be well used for character recognition.