Abstract: It’s difficult for existing Automatic Optical Inspection(AOI) technology to meet TFT online detection requirements of the large-scale high-speed, high precision and high-sensitivity. In view of the problem, hybrid multi-core computer cluster architecture for AOI is put forward and software framework design for AOI system and the involved key software technologies are greatly researched. The research shows that applying fusion programming model for parallel image processing and its implementation strategy which are based on technology such as OpenMP, MPI, OpenCV and IPP, it is effective in overcoming performance bottlenecks that TFT defects detection encountered.

Key words: Automatic Optical Inspection(AOI), thin film transistor, multi-core cluste, fusion programming model, integrated performance primitives

摘要： 针对现有自动光学检测(AOI)难以满足大规模高速、高精度和高灵敏TFT在线检测要求的问题，提出AOI混合多核计算机集群体系架构，研究AOI系统的软件框架设计及其关键软件技术。结果表明，采用基于OpenMP, MPI, OpenCV和IPP等技术的并行图像处理融合编程模型及其实现策略可有效克服TFT缺陷检测的性能瓶颈。

关键词: 自动光学检测, 薄膜晶体管, 多核集群, 融合编程模型, 集成性能原语

CLC Number: 

• TP273