Estimating Intrinsic Images from Image Sequences with Biased Illumination

We present a method for estimating intrinsic images from a fixed-viewpoint image sequence captured under changing illumination directions. Previous work on this problem reduces the influence of shadows on reflectance images, but does not address shading effects which can significantly degrade reflectance image estimation under the typically biased sampling of illumination directions. In this paper, we describe how biased illumination sampling leads to biased estimates of reflectance image derivatives. To avoid the effects of illumination bias, we propose a solution that explicitly models spatial and temporal constraints over the image sequence. With this constraint network, our technique minimizes a regularization function that takes advantage of the biased image derivatives to yield reflectance images less influenced by shading.

• Type: Conference
• Year: 2004
• Where: ECCV
• Authors: Yasuyuki Matsushita, Stephen Lin, Sing Bing Kang, Heung-Yeung Shum
• Keywords: Biased Illumination, Computer Vision, ECCV 2004, Illumination Directions, Reflectance Image Derivatives, Reflectance Image Estimation, Reflectance Images