基于三维场景和图像边缘特征的浅浮雕建模

针对数字浅浮雕设计,通过对已有的浅浮雕建模方法进行分析,提出一种结合深度场边缘特征自动生成浅浮雕的方法。从三维场景中获取深度图,再将深度图分割成块,利用局部图像结构的优点,用随机结构化森林来学习精确且计算有效的边缘检测器,得到的边缘图再进行处理得到平滑且细节增强的浅浮雕轮廓;视边缘连续图为概率场,提出根据概率场对深度场的梯度域进行自适应的非线性压缩;通过求解泊松方程来恢复浅浮雕的高度场。实验结果表明,该方法对于大部分三维场景能在合理时间内自动生成效果良好的浅浮雕模型。

BAS-RELIEF MODELING BASED ON 3D SCENCE AND IMAGE EDGE FEATURES

Aiming at the digital bas-relief design,a method of automatically generating bas-reliefs for depth field edge detection is proposed by the existing bas-relief modeling method.The depth map was obtained from the 3D scene,and then the depth map was divided into blocks.Using the advantages of the local image structure,the random structured forest was used to learn the accurate and computationally valid edge detector.The obtained edge map was processed to obtain a smooth bas-relief contour,and then the edge continuous graph.For the probability field,the nonlinear gradient of the depth field was adaptively compressed according to the probability value.The height field of the bas-relief was restored by solving the Poisson equation.The experimental results show that the proposed algorithm can automatically generate a reasonable bas-relief model for most 3D scenes.