基于小波变换的熔滴过渡图像的边缘检测  

Edge detection of metal transfer image based on wavelet transform

作　　者：杨倩[1] 王广伟[1] 华学明[1] 吴毅雄[1] 

机构地区：[1]上海交通大学材料科学与工程学院激光制造及材料改性重点实验室,上海200030

出　　处：《焊接学报》2007年第4期38-40,共3页Transactions of The China Welding Institution

基　　金：上海市科学技术委员会科研计划资助项目(036105001);国际科技合作重点资助项目:船舶大功率激光建造技术资助项目(2004DFA02400)

摘　　要：利用小波变换和改进的Prewitt算子对GMAW短路过程的熔滴过渡图像进行边缘检测。用中值滤波对采集到的图像做平滑预处理后,使用小波变换将图像分解到各个层次并提取各层的近似系数和细节系数,将各层的对角细节系数加强适当的倍数后重构图像,使用改进的Prewitt算法对重构图像进行边缘检测。结果表明,此方法可以有效地提取真实熔滴轮廓和熔池边缘,小波分解的层数和加强倍数的大小是影响图像边缘提取效果的重要因素,层数和倍数应根据拍摄到的图像的实际情况确定。Wavelet transform and improved Prewitt operator were used to detect the edge of the droplet image in GMAW.After preprocessing,wavelet transform was used to perform decomposition to different detailed levels,and enhance diagonal detail coefficients by multiplying it with the proper parameters,then reconstruct image.Improved Prewitt operation was used to detect the edge of reconstructed image.The experiment results demonstrate that the method is effective to detect the real outline of droplet and real edge of weld pool,and the levels of decomposition and the multiples are vital factors,which depends on the real image.

关 键 词：小波变换 改进的Prewitt算子 熔滴过渡 图像处理 边缘检测 

分 类 号：TG409[金属学及工艺—焊接]