摘要
目的:显微镜过小的景深会导致观测LED芯片时存在模糊的区域,为得到清晰的LED芯片成像图,特开展了相应的图像融合算法研究。方法:通过微动平台不断改变LED芯片与显微镜物镜的距离并拍摄多张聚焦在不同位置的图像,再将得到的图像通过拉普拉斯金字塔算法进行融合。结果:相较于其它图像融合算法,通过拉普拉斯金字塔算法得到的LED芯片融合图最为清晰,标准差可达89.37,平均梯度可达7.76。结论:通过该景深合成系统,实现了图像采样的自动化,解决了显微镜过小的景深而成像模糊的问题,并且融合后的数字图像易于观测和保存。
Aims:To solve the problem of blurring image due to the limited depth-of-field of microscope when observing the LED chip,the corresponding image fusion algorithms were studied.Methods:In the process,the distance between the LED chip and the microscope was changed continuously by the micro motion platform.Multiple LED chip images were captured and were then fused into a clear image through the Laplacian pyramid algorithm.Results:Compared with other image fusion algorithms,the LED chip image obtained by the Laplace pyramid algorithm was the clearest,with a standard deviation of 89.37 and an average gradient of 7.76.Conclusions:Through the super depth-of-field system,the automation of image capturing is realized and solved.The fused digital image is easy to observe and save.
作者
王笔神
刘红林
WANG Bishen;LIU Honglin(College of Optical and Electronic Technology,China Jiliang University,Hangzhou 310018,China)
出处
《中国计量大学学报》
2022年第4期533-538,共6页
Journal of China University of Metrology
