In the process of laser deposition manufacturing, the complex thermal evolution such as heat accumulation and thermal cycle will lead to the overall bending deformation of the substrate and the parts. In order to avoid the decrease of dimensional accuracy of parts caused by excessive deformation, it is particularly important to measure the deformation caused by machining. In this thesis, a binocular vision deformation measurement system is built, and the image segmentation technology is used to extract the target area to filter out the interference of the additive manufacturing background. The combination of image segmentation and various stereo matching algorithms is studied, and the Census algorithm is used to obtain the disparity map of the segmented image. Finally, the singular value decomposition method is used to fit the reference plane to obtain the deformation value. Experiments show that the measurement system can realize the measurement of interlayer deformation. The method proposed in this paper improves the measurement efficiency while ensuring the measurement accuracy.