基于感知组织的图象2D推理和3D解释

本文以室内环境存在的大量结构化物体为背景，研究了从单幅透视图象中恢复３Ｄ结构的感知组织和不确定性推理方法，强调感知组织必须结合３Ｄ解释和验证才能更加有效，本文提出的３Ｄ感知组织方法分三个步骤：（ｌ）２Ｄ图象结构特征可能性推理；（２）基于正交约束的３Ｄ解释及合理性初步验证；（３）基于物体模型的一致性进一步验证，真实景物的图象实验结果说明本文方法在机器视觉中的可行性和有效性。

2D Image Reasoning and 3D Interpretation Based on Perceptual Organization

In this paper a systematic approach of 2D reasoning and 3D interpretation from a single image is presented from the point of view of perceptual organization and uncertainty reasoning.Three steps are involved in the 3D perceptural organization approach : (1)2D grouping and uncertainty reasoning;(2) 3D interpretation and justification; (3) model-based consistency verification. Experiments with real images of indoor scene have shown the robustness and efficiency of the approach.