动车组故障轨边图像自动检测系统图像对比算法研究

Research on image contrasting algorithm for TEDS

动车组故障轨边图像自动检测系统(TEDS)采用模块化、小型化的设计理念,运用图像自动识别技术,对异常图像进行分级报警。在图像自动识别技术基础上,设计了整个图像识别框架,针对不同部件,采用逐级细化、图像对比、分级预报算法,提高动车组故障自动识别效果,达到分级报警的要求。

The modular and miniaturization design concept was adopted in TEDS by using the automatic identification technology to alarm hierarchically in abnormal image. On the basis of automatic identification image technology, the whole image identification framework was designed. According to the different components, the algorithm of stepwise refinement, image contrasting and classification prediction were used to improve the effect of automatic identification for the trouble image of EMU, reach the hierarchical alarm requirements.