Complete 3-D reconstruction of dental cast shape using perceptual grouping

To achieve the complete three-dimensional (3-D) data retrieval of the shape of dentition, dental casts were measured from four directions; occlusal, r ight, left, and labial sides using a line laser scanner. Reconstruction of the entire shape, including undercuts and tooth crowding area, was attempte d by applying a perceptual grouping algorithm, which is one of pattern-reco gnition theories. In the data measured from occlusal, right and left sides, the rows of measu rements were parallel to the frontal plane, and three-directionally combine d data (3-DC data) was accomplished by affine transformation. While, in the labial side, transformation to the frontal plane was done since rows of th e measured data were parallel to the sagittal plane. To combine the labial data with the 3-DC data and reconstruct the complete image, rearrangement o f the order of the data in the file was attempted by applying the perceptua l grouping. That is, the minimum total length of data combining was examine d by considering the factor of proximity and continuity between the data. T he most appropriate order of data combining and recognition of islands were accomplished. Using a computer graphic (CG) with a wire-frame model, complicated regions such as anterior segments showing tooth crowding and undercut area were fou nd to be successfully reconstructed without any data defects. The accuracy of reconstruction was ascertained by comparing the characteris tic distances between apexes of molars in the reconstructed model with the real cast. The difference was within 0.3 mm, and present method for dental cast reconstruction is considered to be satisfactory for the present purpos e such as orthodontics.