在图像分割方法中,CV模型可以得到较好的分割结果,但是模型的收敛速度慢.在三维CV模型检测工件裂纹面的过程中,由于三维CT图像数据量比较庞大且三维CV模型本身分割速度慢,使得检测时间比较长.对于这一问题,研究了一种自适应预处理算法.该算法先对体数据进行三个方向投影,再对投影图利用迭代求最佳阈值的阈值分割方法和自适应矩形框来定位缺陷的大致区域.该方法能够自动适应裂纹面形状变化,同时大幅度减少了需要三维CV模型分割的数据量,可以非常明显地提高分割的速度.实验结果表明利用该预处理算法,三维CV模型的分割裂纹面的速度提高了近8倍.%In digital image processing, C-V model which applies in image segmentation can get superior result, but this model costs too much time. When using 3-D C-V model to detect the crack surface in three-dimensional industrial CT image, the problems are the slow convergence of the model and the large volume data of the industrial CT image. To this question, this paper researched a kind of preprocessing algorithm. This algorithm used adaptive threshold segmentation, which used iteration to get the threshold, and adaptive rectangle on the projection from three directions, which could locate the position of the crack. It also could adapt the variation of the crack and especially reduce the data volume. The experiments show that the pre-processing algorithm can obtain the speedup of almost 8 times compared to the original model.
展开▼
