基于图像分割的多CCD测量系统干扰去除方法

Image interference elimination method based on image segmentation in a multi-CCD measurement system

针对多电荷耦合器件(CCD)光切法物体三维形貌测量系统中杂散光干扰难以去除的问题,提出一种基于图像分割的干扰去除方法,在灰度空间进行阈值分割的同时,在图像中进行区域分割。对干扰源的分析显示,激光条纹关于玻璃底板的反射是干扰的主要来源。玻璃底板以上的部分包含条纹信息,玻璃底板以下为干扰信息存在区域。将玻璃底板在图像中的位置提取出来,利用Hough变换进行检测,求出直线方程。用该直线对测量过程中的图像进行区域划分,有效区域保留原始信息,无效区域置为零,彻底去除干扰。该方法与传统的图像滤波去除干扰的方法相比,可以在去除干扰的同时保持激光条纹的原始信息和测量数据的完整性,该方法已在多CCD三维形貌扫描仪中得到实际应用。

Abstract： Strayed light in 3 D measurement systems based on the multi-charge-coupled device （CCD） light sectioning method is difficult to eliminate. This paper presents an image segmentation based method to eliminate this interference, which simultaneously uses level setting and region segmentation. Analysis of the interference sources reveals that the reflectiofi of light stripes on the glass plate is the main source. Regions over the glass plate are defined as the regions of the light stripe, with the regions below the glass plate having the interferences accordingly. A line equation ~or determining the exact position of the glass plate in the image is deduced using the Hough transform. The line is set as the boundary for the image segmentation with the pixels above the line are considered to be meaningful regions and those below the line set to zero to eliminate the interference. This method more effectively eliminates the interference than the conventional image filtering method for interference elimination while keeping the original light stripe information as well as all the measured data. This has been successfully applied in a multi-CCD 3-D measurement system.