基于高光消除的SFS三维重构算法研究  

Research of Result in SFS Based on Image Highlight Removal

作　　者：李想[1] 胡剑凌[1] 张霞[1] 陈昊亮 

机构地区：[1]苏州大学电子信息学院,江苏苏州215006

出　　处：《电视技术》2015年第21期107-110,共4页Video Engineering

基　　金：国家自然科学基金项目(61271360)

摘　　要：从明暗恢复形状(SFS)是三维重构的有效技术之一。在对SFS算法分析的基础上,针对SFS算法易受高光分量影响的缺点,设计了基于多项式拟合的高光消除方法对输入图像进行预处理,然后对预处理后的图像采用Oren-Nayar光照模型建立辐照度方程,并利用Tsai的线性近似方法对辐照度方程进行求解,进而获得高度信息以实现物体的三维重构。实验结果表明,所提出的方法能较好地消除输入图像中高光分量的不利影响,有效提升三维重构效果。Shape From Shading （SFS） is one of the effective technologies in 3D reconstruction domain. Based on analyzing and considering the drawback of the SFS algorithm which is easily influenced by the highlight area, a polynomial fitting method is introduced as the preprocessing in the reconstruction framework to reduce the influence of highlight area. Then the Oren-Nayar model is exploited to describe and establish the radiation equation. Tsai＇ s linear approximation method is selected to solve the radiation equation, and the height information of the image is retrieved to reconstruct the 3D shape. Experimental results show that the proposed method can reduce the influence of highlight area and promote the result of 3D reconstruction effectively.

关 键 词：从明暗恢复形状 高光移除 光照模型 线性近似 三维重构 

分 类 号：TN911.73[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]