Dichromatic Based Rendering of Texture Images with High Color Fidelity

In this study, an algorithm was developed for the rendering of texture images with high color fidelity. The algorithm is based on the dichromatic reflection model, which recovers the implicit geometrical information of each pixel position in the texture image by exploiting the interaction between object surface and light. Using the recovered implicit geometrical information and a target color, a new texture image can then be synthesized. The synthesized image was then further modified to give correct texture strength. The algorithm developed in this study can be used for both color and gray texture images. It is suitable for the applica- tions in texture simulation and visualization with the advantage of high color fidelity. 4 in- vestigated the proportionality of the pixel deviation to the mean color in each channel and proposed a computa- tional model for the color mapping on texture images. Their method performed well on regular and homoge- neous texture images, but the assumption of the propor- tionality may fail for textures with high contrast. In the study of the texture effect on suprathreshold lightness tolerances, Montag and Berns presented a method of tex- ture image simulation. 5 They first produced a decorrelated color space (DCS) using singular value de- composition (SVD), then modified the mean color of the image, and finally inverted the color transform to gener- ate new texture images. It was noticed that the DCS pro- duced using singular value decomposition is image related, which means that the optimal DCS for the origi- nal image may not be the optimal for the reproduced images with other colors. In fact, the spatial color distri- bution of texture images is attributed to the geometrical structure of the texture samples. Although the previous