【摘要】 还原和恢复变形图像是利用CT扫描非破坏性地研究木材内部水分迁移的关键基础问题。为恢复和还原干燥过程中由于干缩引起的木材CT图像的变形和由于扫描定位误差引起的图像移位,本文选择了三种方法,即图像像素移位法、最小方差法和最大重合法比较其效果。研究表明,像素移位法不适合于干缩较大的木材CT图像的处理,最大重合法的误差比像素移位法和最小方差法小。但对于干燥不均匀、干缩不一致或木材发生翘曲等缺陷的图像,最大重合法也会产生较大的误差。更多还原

【Abstract】 Restoration of deformed images to their original size, shape and location is the key to the utilization of computed tomography (CT) for non-destructive measurement of moisture transfer during wood drying. This paper compares three methods to restore deformed images: pixel displacement method (PDM) that assumes a deformed image has an equal cumulative mass to its original image, least square method (LSM) and most overlap method (MOLM). Both LSM and MOLM apply simple translation and rotation of images but using different criteria to f ind parameters with the best match. By visual observation and pixel location analysis, the most overlap method shows the least error, while the pixel displacement method is not suitable for restoration of images from wood samples with moderate to sever shrinkage. The most overlap method is not suitable for images from unevenly dried or severely distorted wood samples.更多还原