Mesh-guided texture replacement using intrinsic images

In this paper, we present a novel mesh-guided texture replacement tool using intrinsic images. Our system begins with indicating the region of interest by the user´s interactive segmentation, then decomposes the segmented region into the reflectance and illumination components using intrinsic image recovery method. After that, we generate a triangle mesh to guide the computation of texture coordinates with the interest region. The texture coordinates for retexturing are further optimized. Finally, when the target sample texture is specified, the retexturing image is created by multiplying the modified reflectance component by the original illumination component within the replaced region. As demonstrated later, our mesh-guided approach can produce high quality texture editing results such as image retexturing and texture transfer, while the lighting and shading effect on the original image is well preserved.