Shape from Texture using Locally Scaled Point Processes

Shape from texture refers to the extraction of 3D information from 2D images with irregular texture. This paper introduces a statistical framework to learn shape from texture where convex texture elements in a 2D image are represented through a point process. In a first step, the 2D image is preprocessed to generate a probability map corresponding to an estimate of the unnormalized intensity of the latent point process underlying the texture elements. The latent point process is subsequently inferred from the probability map in a non-parametric, model free manner. Finally, the 3D information is extracted from the point pattern by applying a locally scaled point process model where the local scaling function represents the deformation caused by the projection of a 3D surface onto a 2D image.

💡 Research Summary

The paper presents a novel statistical framework for recovering three‑dimensional surface orientation from a single two‑dimensional image containing near‑regular texture. Unlike earlier shape‑from‑texture approaches that rely on strong assumptions such as texture homogeneity, isotropy, stationarity, or orthographic projection, the authors model texture elements as realizations of a latent point process with only a mild convexity requirement.

The methodology consists of three distinct stages. First, image preprocessing creates a probability map Y(x)∈