Compared to conventional dental panoramic radiograph, synthesizing panoramic radiograph directly computed from dental cone-beam computed tomography (CBCT) images provides high-resolution images with reduced geometric distortion. Previous approaches required manual operations or were based on CBCT images without metal artifacts or missing teeth. In this paper, we introduced a fullyautomatic method that can be applied to CBCT scans under severe conditions, which consist three main steps: bone and teeth segmentation, mandible curve extraction, and panoramic mapping. An adaptive method was first used for bone and teeth segmentation. Then the internal mandible curve, which provides a stable curve regardless of metal artifacts or missing teeth, was extracted and utilized for panoramic mapping. Our synthesizing method showed stable performance on various CBCT data sets.