By dividing fabric image into many windows with the same size , choosing a window with minimum local entropy as a region of interest , thresholding the region and filtering out noise through opening function of mathematical morphology , calculating the defect shape factors as recognition parameters , a algorithm and method of detection of defects in fabric is studied : which has advantages of high identification , correctness , and fast inspection speed mainly because it can avoid complicated calculation of whole image and global searching when the feature parameters are extracted 摘要将织物图像分成大小相同的局部窗口，选取局部熵最小的窗口为待研究的感兴趣区域，在此区域内分割出疵点图像并用数学形态学中的开运算滤除噪声，计算疵点形状因子等作为识别参数，研究表明此方法因能避免对整幅图像进行复杂运算，在提取特徵参数时对图像的全局搜索，具有识别正确率高、检测速度快等优点。
The present research status and some methodologies about the 3d reconstruction and visualization are introduced in this paper , while three important contents , i . e . , image pre - processing , 3d reconstruction and 3d reveal are more concerned . image pre - processing includes interpolating , filtering , gray adjusting , rotating , zooming , selecting voi and reducing elements which are very important for the next step . the work of this paper use marching cubes ( mc ) algorithm to reconstruct the 3d object after analyzing the volume rendering method and the surface rendering method 在二维图像预处理方面讨论了滤波、灰度修正、旋转和缩放、感兴趣区域voi （ volumeofinterest ）的提取、减少体数据集中象素的数目、层间插值等，这些处理可以减少噪声的干扰，使物体的特征更加突出、需要处理的数据大大减少、节省存储空间、加快绘制速度，这些处理对于后一步的三维重建是非常必要的；在三维重建方面，分析了面绘制和体绘制各自的优点和缺点，选取了面绘制中经典的mc算法实现三维重建；在三维显示中实现对体数据的显示、剥离显示、旋转、等值面显示、沿x 、 y轴的剖切显示、光照显示等效果。
