烟草异物实时图像敏感技术研究

介绍了烟草异物实时图像检测系统的工作原理,详细分析了系统中的实时图像敏感技术。首先分析系统中上位机和TMS320DM642型数字信号处理(DSP)处理器之间的实时图像数据传输原理。包括DM642EVM的外设部件互连标准接口(PCI)数据通信原理、主机与DSP实时通信的具体实现方式、使用增强型直接内存存取(EDMA)实现DSP内部数据传输的方法。其后介绍了烟草异物图像实时识别的原理,设计了基于修正灰度值IE、色调值H和灰度1阶中心矩D1等3个特征量的烟草异物实时图像检测算法,并以实验结果表明了检测算法的有效性。

Research on Real-time Sensing Technique of Tobacco and Foreign Material Image

In this paper,the working principle of real-time image detecting system is introduced,and the real-time image sensing technique of the system is analyzed in detail.At first,principle of real-time data communication between superior PC and DSP processor of TMS320DM642 is analyzed.This includes the following three part: Data communication principle of DM642EVM PCI bus;Concrete implementation ways of real-time communication between host and DSP processor;Methods Used to realize the DSP interior data transmission by EDMA.Then the principle of real-time image recognition between tobacco and foreign material is introduced,and the real-time image detecting algorithms between tobacco and foreign based on modified gray-scale IE,hue H and first-order center moment of gray-scale D1 is designed.Validity of this algorithms can be verified by the experimental result.