Automatic Shade Matching in Dentistry Using Digital Cameras

Methods: Vitapan-3D shade guide and Nikon995 digital camera are employed for experiments. Shade tab images are captured at different time intervals under clinical environment. Content-based images are cropped from shade tab excluding proximal shadow and cervical root region; each image is divided into 10 x 2 blocks. Features from different color spaces are being analyzed. Algorithms measuring blocks differences between shade colors have been established and compared. Different sampling methods are discussed. Effective matches are defined when the least n differences between the shade tabs are attained.

Results: With our proposed Sa*b* L1 norm metrics, the top one matching rate is 0.62 under open conditions. This rate is higher than the previous research reported about 0.50 by computerized match in limited conditions. The S feature is defined in HSV color space, a* and b* features are defined in Lab color space. The top five matching rate is 0.92, which is helpful in clinical applications.

Conclusions: Measurements counting for color gradation along the contents will have better results. This means that spatial relations within the color contents play an important role in shade matching. The sample mean is less useful because the spatial relationship of the features cannot be averaged. ΔΕ is shown to be an impropriate measurement. In addition, Sa*b* are suitable features for shade matching in clinical conditions using digital cameras.