摘要
镁熔液图像背景十分复杂,为了检测出镁熔液图像中的弱小气泡,本文提出了一种基于最小一乘准则和区域生长分割的检测算法。该算法首先运用最小一乘准则和遗传算法对镁熔液图像进行背景预处理,再运用区域生长对检测到镁熔液图像中的气泡目标进行分割。试验证明,该方法能够成功地检测出弱小气泡。
A detection algorithm was presented for weak and dim bubble in magnesium melt based on the least absolute deviation and region growing segmentation,which aimed at image identification of the first bubble under complex background.At first,a background prediction model of magnesium melt based on the least absolute deviation was established.The extreme value was extracted by the genetic algorithm to predict the background.Then,the residual image was got by subtraction between original image and the prediction image.And then,the frame difference was used for the two residual images.Finally,the first bubble in magnesium melt was detected by region growing segmentation.The experiments show that this method can successfully detect the weak and dim bubble.
出处
《铸造》
CAS
CSCD
北大核心
2014年第11期1134-1137,共4页
Foundry
基金
国家自然科学基金(51374007)
关键词
镁熔液
第一气泡
最小一乘
遗传算法
区域生长
magnesium melt
weak and dim bubble
least absolute deviation
genetic algorithm
region growing segmentation
