中文摘要：

在条纹投射测量方法中，由于三角测量法固有的特点，将有效测量区域从含有背景、阴影、表面不连续等区域的条纹图中分辨出来，完成相位去包裹计算，是一个实际测量中的难点之一。本文借鉴图像分割的原理，对调制度分布图进行分析，提出一种基于调制度自动分割的物体轮廓有效测量区域自动识别方法，实验结果表明，该方法能够实现阈值的最佳选取，自动识别物体轮廓有效测量区域，并根据变形图与极点堂关系，提出由A trous algorithm of wavelet transform（多孔小波算法）来检测曲面变形的奇异点，通过奇异点辅助来连接截断线，从而实现快速有效地解相并提高三维测量的精度。

英文摘要：

Using the method of stripe casting, phase unwrapping is a difficulty in the practical measurement because of the original shortcoming that caused shadow, background and discontinuous field mixed in the useful information. This thesis adopted the theory of image segmentation, put forward an effective automatic recognition method of the object profile. This way can get the best threshold to recognize the surface and get a new way to line the branch-cut , achieving an efficient and fast unwrapping method and improving 3D measurement accuracy.