基于个数判断边缘的镁熔液第一气泡图像检测

摘要

An adaptive background prediction based on number-judgment edge is proposed for image identification of the first bubble on magnesium melt surface. In this algorithm, number-judgment edge method is used to determine the current pixel's location, and then different prediction algorithms were adaptively selected for processing pixels of different zone. If the current pixel exists in the edge zone, the maximize background prediction is adopted, otherwise not processed, at last the first bubble was detected by frame difference and threshold segmentation. Further, the algorithm was compared with the conventional maximize background prediction algorithm. The results show that this algorithm can detect the location of the first bubble accurately, and suppress the background more effectively and improve the signal to noise ratio more significantly.%针对镁熔液表面第一气泡的图像识别,提出了基于个数判断边缘的自适应背景预测算法.该算法首先利用个数判断边缘的方法判断当前像素点所处的位置,然后对不同位置的点自适应地采用不同的预处理算法,若当前像素点处于边缘区域则进行最大化背景预测处理,若与前像素点处于非边缘区域则不做任何处理保留其原始灰度值,最后经帧间差分和阈值分割,将第一气泡检测出来,并与传统最大化背景预测预处理算法进行了对比.结果表明,该算法不仅能准确地检测到第一气泡,而且还能更有效地抑制背景,信噪比明显提高.