一种基于遗传算法的明暗恢复形状重构模型光顺方法.docVIP

  一种基于遗传算法的明暗恢复形状重构模 型光顺方法# 王浩臣，王军，吴凤和，李秀亮** 5 10 （燕山大学机械工程学院，河北秦皇岛，066004） 摘要：针对 SFS 重构曲面光顺性差的问题，提出基于遗传算法的明暗恢复形状重构模型光 顺方法。首先，利用 SFS 方法获取曲面的三维点云数据；再通过 B 样条插值方法建立重构 曲面的数学模型，并通过控制顶点的反算重构曲面；最后，根据 SFS 重构曲面的特点确定 曲面光顺准则，并选择合适的编码方法、适应度函数及遗传操作方法去除坏点，最终实现基 于遗传算法的 SFS 重构模型光顺。实验结果表明：该方法能够有效去除重构曲线及曲面上 的坏点，提高光顺处理效率，保证 SFS 重构曲面的形状精度。 关键词：曲面重构；光顺；明暗恢复形状；遗传算法；B 样条 中图分类号：TP391.41 15 A Smoothing method of Shape from Shading Reconstructed model Based on Genetic Algorithm WANG Haochen, WANG Jun, WU Fenghe, LI Xiuliang (Yanshan University, College of Mechanical Engineering, Qinhuangdao, Hebei,066004) 20 25 30 Abstract: To improve the smoothess of SFS reconstructed surface, a genetic algorithm based smoothing method of SFS reconstructed model was proposed. Firstly, three-dimensional point cloud data were acquired by using Shape from shading method. And then, the mathematical model of reconstructed surface was established by B-Splined interpolation method, and the parameters of the reconstructed surface were determined through the reverse calculating of control points. Finally, the evaluation criteria of smoothness for curves and surfaces were given according to the features of SFS reconstructed surface, and the bad points were removed by using the genetic algorithm method with suitable encoding method, fitness function, and genetic operations, so the reconstructed surface was smoothed. Experiments results show that the method proposed can effectively remove the bad points on reconstructed curves and surfaces, improve the efficiency of smoothing, and maintain the shape precision of SFS reconstructed surface. Key words: Surface reconstruction; Smoothing; Shape from Shading; Genetic algorithm; B-spline 0 引言 近年来，随着计算机视觉和图像处理技术的快速发展，明暗恢复形状方法(Shape from 35 Shading,简称 SFS) [1-2] 程、人脸识别、自然景物模拟等[3-4]领域，具有十分广阔的应用前景。但在曲面重构过程中， 由于图像质量的影响以及 SFS 方法自身存在的缺陷，使得重构曲面存在坏点，既不满足产 品的设计要求，也不便于加工。因此，必须对 SFS 重构模型进行光顺处理[5-6]，以满足实际 应用要求。 40 根据曲面光顺处理[6-7]时修改区域的大小，光顺处理方法主要可分为局部光顺法和整体 光顺法两大类。局部光顺法每次处理时只修