摘要
在基于条纹投影和相位分析的三维面形测量中,由于被测物体表面标志点或复杂面形的阴影遮挡存在,会造成变形条纹局部区域的条纹数据缺失,影响相位和高度信息的最终重建,需要人为地对缺失图像信息进行修复。提出了一种新的缺失条纹数据修复方法——基于模版匹配的图像修复算法,通过图像中已有条纹信息(特别是与待修复区域周围相位信息相似度较高的已知条纹信息)对缺失的变形条纹信息进行估算,实现数据修复。该方法修复效果好,运算过程无需人为参与,便于计算机自动实现,尤其适合于待修复图像整体结构明显、纹理清晰图像的数据修复,有助于提高被测物体相位计算质量和在此基础上的三维面形重建质量。
In optical three-dimensional（3D） measurement with structured illumination, some data would be missed in the deformed fringe pattern caused by the attached marks or shadowing of tested object＇s protrusions. A novel method of image restoration -- exemplar-based fringe inpainting algorithm is presented to interpolate the missing data in deformed fringe pattern. It uses the known fringe information with highly similarity around the missing area to inpaint the missing data. This new algorithm gains well restored result especially when it is used to cope with the image with clear structure and texture.Without any predefined parameters in the inpainting process, this exemplar-based fringe inpainting algorithm could be used in missing phase information interpolation for optical 3D measurement application in many fields.
出处
《光学与光电技术》
2014年第3期24-30,共7页
Optics & Optoelectronic Technology
基金
教育部新世纪优秀人才支撑计划(NCET-11-0357)
四川省学术和技术带头人后备人选培养基金资助项目
关键词
光学测量
图像修复
模版匹配
数据插补
标志点
阴影遮挡
optical measurement
image inpainting
exemplar-based
data interpolation
marked point
shadow
