Quantitative quality measure for photorealistic three dimensional models

The demand for virtual photorealistic models has been recently increased especially in the field of documenting large archaeological sites. These models are commonly assessed qualitatively as no quantitative measure tool is available. In this work, a new approach is presented for a quantitative measure of models texture similarity. Deciding for the reference texture is crucial. Metrics that have been used to assess image stitching quality in panoramic images are employed in the presented approach. A proposed metric is also presented and employed to assess the overall quality of the whole model’s texture. The work is then extended to manage misalignments that might appear in textured models in case of using wide angle lenses. A function that fits a second order polynomial equation is derived experimentally. The approach has been validated by three real models. The quantitative measure approach and the visual inspection of subjects for the models show a consistent correlation between the results.