摘要
采用自行设计的三光路传感器,获取了多方向上清晰的填丝脉冲铝合金熔池图像,并对斜后方图像进行了处理.针对斜后方图像特点,采用退化恢复、滤波、边缘检测、自动阈值分割、投影细化方法进行了预处理;对于预处理后的图像噪声,经过阈值滤波去除,获得间断熔池边缘图像;最后采用任意位置椭圆来拟合熔池边缘,拟合结果可以较好地反映熔池形状.
Three-direction visual sensor was designed and clear multi-direction image was got in aluminum alloy pulsed gas tungsten arc welding（GTAW） wire feeding. This paper mainly processed the back-topside image. Considering the characters, series of image technology including degraded recover, filter, edge de- tection, automotive threshold segmentation, projecting thinning are used to do preprocess work. For the noise in preprocessed image, size filter is used to delete. Then discontinuous edge is left. Lastly, ellipse is used to fit the weld pool edge. The results fit the pool very well.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2008年第1期42-45,共4页
Journal of Shanghai Jiaotong University
基金
国家自然科学基金资助项目(50575144)
上海市科委“登山计划”重点基础研究项目(06JC14036)
关键词
铝合金
视觉传感
图像处理
aluminum alloy
visual sensing
image process
