【摘要】 针对纺纱过程中手工检测毛条直径时间长,速度慢,精度不高等问题,设计了一种基于ARM嵌入式图像处理技术的毛条均匀度在线检测装置。首先,在ARM嵌入式系统内移植开源视觉库(OpenCV),使用OpenCV作为检测装置的核心算法库和数据处理库函数;其次,在预处理阶段,利用分段构造线性滤波函数替代复杂的非线性滤波函数,以OpenCV集成的方框滤波函数(box filter)重新构架非线性滤波器,滤波时间只有经典双边滤波器的5%;最后,利用交叉编译器在Qt开发环境内生成ARM-Linux环境下的可执行程序,使用USB显微摄像头采集图像,实现毛条直径的在线检测。在Crotex-A8内核的ARM嵌入式内进行测试,单次检测耗时在900 ms以内。实验结果表明,在线监测装置能够在光照不足的条件下自适应检测毛条直径,同时能够对毛条直径进行分类,人为实时在线标定,检测结果与手工测量误差低于5%。更多还原

【Abstract】 The traditional yarn`s diameter inspection process has characteristics of inflexible,long delay,and low accuracy. In order to solve those problems,a yarn inspection instrument based on digital image processing techniques was designed. Firstly,an open source computer vision was transplanted into ARM embedded system,which as the core algorithm library and data processing library. Secondly,in the image pre-processing stage,a method of segment linear filter was given. Using the box filter to rebuild a new non-linear filter to replace the traditional filter method,and the time spent of new was about half of old.Finally, the instrument used microscopic camera to photo picture in the ARM-Linux embedded equipment. The simulation results show that this equipment can automatic work in the different environment,and classify a diversity of yarn efficiently. The average measured accuracy was within 5 percent in the time of 900 ms.更多还原