一种基于网格的从明暗恢复形状方法  

Method for shape from shading based on mesh

作　　者：田丰[1] 高骞[1] 郭巍[1] 仇庆丰[1] 王传云[1] 

机构地区：[1]沈阳航空工业学院计算机学院,沈阳110136

出　　处：《计算机应用》2009年第1期74-77,共4页journal of Computer Applications

摘　　要：针对目前"从明暗恢复形状"时存在计算复杂及收敛性差等问题,提出一种基于网格的解决方法。首先通过图像预处理及区域划分操作将原始图像分割成多个亮度离散的多边形区域,并逐步转换成平面三角形网格,然后根据Lambert定律建立关于空间网格顶点坐标的方程组,在使用最小二乘法求解后,直接将结果应用到OpenGL或DirectX等实时渲染引擎的图形管道中,实现模型的建立与显示。经仿真实验证明该方法计算方便,可逼真重建原始物体的形状。Concerning the current problems of high computation cost and poor convergence performance during Shape From Shading （SFS）, a mesh-based resolution was proposed. The original image was divided into several regions by the means of image preprocessing and region segmentation and were transformed them into triangular mesh. And then equations with regard to each vertex of the mesh in three-dimensional space were established. Finally the result resolved form equations was straightly applied to the graphic pipeline of real-time three-dimensional render engine such as OpenGL or DirectX to accomplish the model reconstruction and rendering. The results of simulation show that the method is of convenient calculation and can reconstruct the original model realistically.

关 键 词：从明暗恢复形状 网格重建 图像处理 Lambert定律 最小二乘法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]