Abstract: Reconstruction speed and accuracy are two important indicators of industrial CT products. According to the MC surface rendering algorithm, an improved algorithm is proposed based on the comparability region segmentation. The accurate segmentation result is used as the input of MC to generate equivalence surfaces. The detection efficiency can be improved significantly. According to ray casting algorithm, an 2D maximum entropy threshold segmentation preprocessing is proposed. The 2D histogram entropy maximization is used to define the extent of threshold, and research the optimal combination of the parameters. The results of reconstruction show that the improved algorithm can greatly reduce reconstruction volume data and improve the efficiency of ray casting.

Key words: industrial Computerized Tomography（CT）, surface rendering, volume rendering, Marching Cube（MC） algorithm, ray casting algorithm

摘要： 工业CT图像的重建速度和精度是工业CT产品的两个重要指标。针对面绘制的MC算法提出了一种基于相似性区域分割的三维工业图像表面重建算法，实现了准确分割，并利用分割结果精确地提取等值面，显著提高了检测效率；针对体绘制的光线投射算法提出了一种基于二维最大熵阈值的分割预处理方法，利用二维直方图熵最大化寻找阈值的最佳组合，能有效减少重建体数据量，实测数据表明体绘制速度明显提高。

关键词: 工业CT, 面绘制, 体绘制, MC算法, 光线投射算法