Purchase this article with an account.
Masataka Sawayama, Shin'ya Nishida; Discrimination of highlights from reflectance changes using isophote maps of surface images. Journal of Vision 2014;14(10):459. doi: https://doi.org/10.1167/14.10.459.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Low-level image statistics such as skewness of luminance/subband histogram are effective cues for surface gloss perception (Motoyoshi, Nishida, Sharan & Adelson, 2007). Nevertheless, these statistics alone cannot explain why specular highlights look more like white blobs produced by surface reflectance changes when the highlights are inconsistent in position and/or orientation with the diffuse shading component (Marlow, Kim & Anderson, 2011). Here we show that specular-shading consistency can be judged by 2D image inspection without referring to the 3D shape or illumination of the surface. A surface image can be decomposed into reflectance, shading and specular images. Whereas the reflectance image is independent of the shading image, the specular image is dependent on the shading image, because the shading intensity is a function of the incident angle of light, while the specular intensity is a function of the incident and viewing angles of the light. As a result, within each highlight region, the surface normal directions are nearly uniform, and so are the shading intensities. Diffuse components under and around veridical highlights should have similar intensities. Violation of this constraint implies a reflectance change, and this can be checked with isophote maps. We made consistent-glossy or inconsistent-blobby images by combining specular and diffuse patterns, rendered from the same or uncorrelated depth profile, respectively. The image analyses showed that for consistent images, addition of the specular component affected little the isophote-contours of the original matte image. For inconsistent images, on the other hand, addition of the specular component significantly altered the isophote-contours of the matte image. This explains why a histogram change can perceptually erase highlights in the consistent images (i.e., making them natural pure matte images), but not white blobs in the inconsistent image. These observations suggest that isophote maps contain useful 2D information for discrimination of highlights from white blobs.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only