The measurement method based on laser-aided machine vision has been widely used in 3D shape measurement for large-scale components and parts at present. To solve the problem of the inaccurate extraction for components′ boundaries, a high-precision method of 3D shape measurement for large-scale parts was proposed in this paper. Firstly, an adaptive light stripe locating method based on time flow was proposed according to the sequence of acquiring images, by which the region of the light stripe can be located in real time. Next, a boundary detection method based on light stripe mutations was presented to acquire the row or column coordinates of the object boundaries in the light stripe. Then the coordinates of the boundaries can be determined precisely combined with the extraction results of light stripe centers. And the effective light stripe centers for measurement within the two determined boundary feature points can be retained. At last, the extracted light stripe centers were matched and reconstructed to accomplish 3D shape measurement. Experiment results show that the measurement accuracy of this method reaches 0.06 mm and precise boundary information of objects can be acquired. The cost of computation is also reduced greatly. Thus, the proposed method can satisfy the measurement requirements of high precision and rapid speed.