基于改进Snake算法的坡口中光致等离子体图像分割

Image Segmentation of Laser Induced Plasma Inside Groove Based on Modified Snake Algorithm

光致等离子体是CO2激光焊的重要现象。厚板激光焊时,坡口中的光致等离子体图像受熔池等噪声影响,传统边缘提取算法易产生伪边缘,给后续分析计算带来困难。设计了坡口中光致等离子体图像采集系统,所获得的图像经高斯滤波后,为了克服熔池对光致等离子云图像的干扰,改进了主动轮廓模型Snake算法,基于OTSU算法完成等离子云图像的二值化和初步分割,设计了Snake模型能量函数并优化了相关参数,提取了坡口中光致等离子体边缘,与传统Sobel算子等方法比较,提高了等离子云边缘提取的准确性和可靠性。

Laser induced plasma is an important phenomenon during high power CO2 laser welding. Plasma image inside groove is affected by molten pool interference for thick steel plate. Pseudo edge extracted by traditional image processing algorithm lead to the failure of further analysis. Specific acquisition system was setup to capture plasma image inside groove. As an emerging active contour model, Snake algorithm was modified to overcome the interference of molten pool. Binary plasma image was obtained by optimized threshold .The otsu algorithm after image was filtered. Energy functions and relevant parameters of Snake model were designed to extract more credible and rational edge of plasma plume. Compared with the traditional methods, the experimental results show that the precision of plasma image segmentation is greatly improved.