Traditional image-based detection methods for pipelines lack quantification information (e.g., depth, area, and perimeter of defects) despite the high accuracy. Furthermore, existing three-dimensional (3D) reconstruction methods based on laser and depth cameras are expensive and not maneuverable. To remedy these problems, a simple and novel measurement system for sewer pipeline potholes based on low-cost 3D reconstruction is proposed. First, a sparse reconstruction method based on structure-from-motion (SFM) is proposed to estimate the camera parameters and reconstruct the sparse point cloud from multi-view images that are easily acquired. Second, a dense reconstruction method based on multi-view stereo (MVS) is proposed to generate the depth maps and dense 3D points that can provide a stereo display of pipelines. Third, an automatic segmentation and measurement method based on cylinder fitting and projection for pipeline potholes is proposed. The measurement information of potholes is obtained by projection, triangulation and boundary search. Furthermore, a metric calibration method is proposed to convert the voxel size to the actual size. Comparison experiments show that the average errors of maximum depths, mean depths, areas and perimeters between the predicted and real values are 9.48 %, 13.16 %, 6.70 % and 14.01 %, respectively. Furthermore, the measurement method is robust to the point density of the potholes. The whole proposed system only requires several overlapping images taken from ordinary cameras, which is a low-cost and accurate way for 3D reconstruction of pipelines and measurement of defects.