Nondestructive and high-throughput screening of 3D scanning method in the early detection of scoliosis has played a major role. Gaussian curvature and mean curvature methods are often used in current research to express the 3D point cloud information of the human body. Meanwhile, Gaussian curvature is often used by scholars to detect the characteristic regions of the human back, such as: the spine-ridge, scapula, C7 and other locations. However, Gaussian curvature does not perform well enough in the task of identifying the spine-ridge area in the shape of a strip and the lumbar fossa and buttocks in the shape of a round pit. Therefore, based on the fact that the difference between the maximum curvature and the minimum curvature values at the spine-ridge is larger than the difference between the other locations on the back, this paper proposes a new method for calculating the curvature. This curvature calculation method is composed of a dynamic Gaussian curvature value that combines the quotient of maximum curvature and minimum curvature. The values at the spine-ridge under this curvature method are larger than those at other sites. This curvature calculation method enhances the contrast at the spine-ridge in the curvature map, making it convenient for spine-ridge localization. The results show that this new 3D curvature calculation method improves the energy gradient index by 512.688%.