基于卷积神经网络的高通量蓝相液晶识别  

High-throughput blue phase liquid crystal recognition based on convolutional neural network

作　　者：张亚倩 崔永丰 王浩[1] 何万里[1] 张磊[1] 杨洲[1] 曹晖[1] 王冬[1] 李宇展 ZHANG Ya-qian;CUI Yong-feng;WANG Hao;HE Wan-li;ZHANG Lei;YANG Zhou;CAO Hui;WANG Dong;LI Yu-zhan(School of Materials Science and Engineering,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]北京科技大学材料科学与工程学院,北京100083

出　　处：《液晶与显示》2022年第8期972-979,共8页Chinese Journal of Liquid Crystals and Displays

基　　金：科技部国家重点研发计划(No.2018YFB0703703);国家自然科学基金(No.61370048,No.51673023,No.51773017,No.51973017)。

摘　　要：随着液晶显示技术的发展,蓝相液晶由于存在诸多优点而进入研究人员的视野,并吸引着人们的持续关注,但是其在研究和实用过程中存在许多困难,例如蓝相只在很窄的温度区间内存在,所以对于蓝相液晶的快速识别和蓝相存在温度区间的快速计算就显得格外重要。本文通过机器学习算法训练得到模型,结合Labview软件可以实现液晶相态的快速识别和蓝相温域的快速读取计算,在实验过程中对159840张样品点相态图像的整体识别准确率在93%以上。本文研究结果可以更高效识别蓝相和计算蓝相温域,更快速筛选合适的蓝相液晶配方,从而提高实验研究效率,推动蓝相液晶早日应用于显示器件。With the development of liquid crystal display technology,blue phase liquid crystal has entered the researcher’s field of vision due to its advantages and has attracted continuous attention.However,there are many difficulties in its research and practical application.For example,the blue phase only exists in a very narrow temperature range,so the rapid identification of the blue phase liquid crystal and the rapid calculation of blue phase temperature range are extremely important.In this paper,a model trained by a machine learning algorithm combined with Labview software can realize the rapid recognition of the liquid crystal phase state and the rapid reading and calculation of the blue phase temperature range.During the experiment,the overall recognition accuracy of 159840 sample point phase images was above 93%.The research results of this paper can help to effectively identify the blue phase,quickly obtain the blue phase temperature range,and quickly screen the applicable blue phase liquid crystal formula,so as to improve the research efficiency of blue phase materials and to promote their application in display devices.

关 键 词：蓝相液晶 机器学习 卷积神经网络 图像识别 

分 类 号：O753.2[理学—晶体学]