基于GA-Otsu法的图像阈值分割及定量识别

发布时间：2018-03-24 20:45

  本文选题：计算机应用　切入点：热波检测　出处：《吉林大学学报(工学版)》2017年03期

【摘要】：针对传统Otsu法在图像分割中存在的计算量大、效率低的问题,提出了一种基于GA-Otsu的热波图像分割方法。该方法充分发挥了遗传算法的全局寻优能力,快速地求解热波图像分割的最佳阈值,缩短了损伤的分割时间。通过与人工阈值法的分割结果进行对比,表明本文方法不仅保持了冲击损伤的基本形状,而且对冲击点位置的分割提取也较为准确。最后,通过对缺陷的定量识别验证了本文方法在热波检测图像分割中的有效性。
[Abstract]:Aiming at the problems of large computation and low efficiency of traditional Otsu method in image segmentation, a thermal wave image segmentation method based on GA-Otsu is proposed. This method gives full play to the global optimization ability of genetic algorithm. The optimal threshold of thermal wave image segmentation is solved quickly and the time of damage segmentation is shortened. By comparing with the segmentation result of artificial threshold method, it is shown that the proposed method not only keeps the basic shape of impact damage, And the location of impact point segmentation is also more accurate. Finally, through the quantitative identification of defects, the effectiveness of this method in thermal wave detection image segmentation is verified.
【作者单位】： 西北大学信息科学与技术学院;咸阳师范学院教育科学学院;北京师范大学信息科学与技术学院;
【基金】：国家自然科学基金项目(61373117,61305032)
【分类号】：TP391.41
，

本文编号：1659957