Project Image Modeling Smoothly Varying Texture
Utilizing the steerable pyramid of Simoncelli and Freeman [14] as a basis, we decompose textured regions of natural images into explicit local attributes of contrast, bias, scale, and orientation. Additionally, we impose smoothness on these attributes via Markov random fields. The combination allows for demonstrable improvements in common scene analysis applications including unsupervised segmentation, reflectance and shading estimation, and estimation of the radiometric response function from a single image.

People Involved: Jason Chang, John W. Fisher III
The following shows the estimated parameters. The last image shows the features with the smooth fields removed, which we call the intrinsic texture image.
Original Image Contrast Energy Bias (Average Intensity Orientation Scale (Brighter is coarser) Intrinsic Texture Image
Using these features, we can perform image segmentation, estimation of the radiometric response of a camera, and shading/reflectance decomposition. The following examples illustrate the algorithm
Image Segmentation Shading Reflectance Shape View 1 Shape View 2

The Entire Process

Joint segmentation, radiometric response estimation, shading estimation, and shape from shading. The top left image is the current segmentation. The top middle image is the labels assigned. The top right image represents the likelihood of any pixel belonging to a specific region label. The bottom images represent the estimated shape of the zebra based on the current segmentation.

Refereed Conference Papers
1 result
[1]J. Chang, J. W. F. III, "Analysis of Orientation and Scale in Smoothly Varying Textures", in IEEE International Conference on Computer Vision, 2009. [bibtex]