"A Bayesian Approach for Shape Extraction & Scene Segmentation of 3D Textured Scenes From Image Data"

本项目将为透视投影下的马尔可夫纹理模型的理论研究做出贡献。科恩博士正在扩展他在基于马尔可夫的纹理识别和分割方面的工作，将3D透视扭曲包括在内，从而开发出纹理分析的理论工具，并建立了一种基准技术，可以与其他透视纹理建模方法进行比较。马尔可夫纹理模型在机器视觉和计算机图形学中可能特别有用。

This project will contribute to the theory of Markov texture models under perspective projection. Dr. Cohen is extending his work on Markov-based texture identification and segmentation to include 3-D perspective distortions, thus developing theoretical tools for texture analysis and establishing a baseline technique to which other perspective texture modeling approaches can be compared. The Markov texture models may be of particular use in machine vision and in computer graphics.