【作者】 郑晓悦；

【导师】 余辉；

【作者基本信息】 天津大学 ， 生物医学工程， 2012， 硕士

【摘要】 菌落检测是农业、食品、医药卫生等行业进行质量检测的重要方法之一。依靠肉眼观察的人工检测方法速度慢、结果重复性差,而且工作繁重乏味,工作效率低。目前,市场上虽然出现了菌落自动计数仪,但大都以PC机作为运算平台,便携性差,难以满足实时检测要求。近年来嵌入式技术飞速发展,嵌入式设备在运算速度和存储容量上都有了长足的进步,嵌入式操作系统的发展也日趋成熟。由于嵌入式设备本身具有便携性好、成本低的特点,使嵌入式技术得到了更为广泛的应用。针对菌落检测对实时性和便携性的要求,本研究结合嵌入式技术,设计了基于ARM平台的嵌入式菌落智能计数系统,并完成了原型样机的开发。系统以S3C6410作为核心处理器,采用广角红外CCD摄像头作为视频采集设备。为了提高嵌入式菌落智能计数系统的整体运行效率,我们对系统使用的Windows CE 6.0内核进行了裁剪和定制;为了达到最好的检测效果,本设计对样机结构和照明系统进行了精心设计。菌落自动计数算法是整个系统性能优劣的关键。我们使用OPENCV计算机视觉库开发了菌落智能计数算法,研究了一种基于霍夫变换的培养皿边缘提取算法,取得了良好的效果。粘连菌落的分割是菌落自动计数算法的难点,本文提出了一种基于迭代腐蚀算法的菌落计数算法,并通过多次条件膨胀的方法消除了传统迭代腐蚀算法对较大菌落的“过分割”问题,使平均计数偏差控制在6%以内,取得了很好的效果。最后,本文完成了算法的移植和PC演示版软件的开发,并最终实现了便携式菌落智能计数系统的开发和测试工作;总之,本研究对菌落自动计数算法进行了有意义的探索和尝试,并在此基础上开发了嵌入式样机系统,研究成果具有一定的理论意义和应用价值。更多还原

【Abstract】 Colony detection is an important method of quality inspection in agriculture, food and medicine industries. But it is a hard and boring work to detect colony using the method of traditional visual check. Currently, most of the colony automatic counter is developed based on PC platform which has poor portability and could not meet the requirements of field detection. With the rapid development of embedded technology, the embedded device which has the advantages of low cost and good portability has been used more and more widely.An embedded intelligent colony counting system based on ARM was designed in this paper and a kind of embedded colony automatic counter prototype was developed. The S3C6410 chip was used as the CPU of the system. The wide-angle infrared camera was chosen for colony image acquisition and a customized Windows CE 6.0 kernel was designed to improve system efficiency. Prototype structure and the lighting system were well designed for the best results.In this paper, an intelligent automatic colony counting algorithm was developed with the OPENCV computer vision library and was tested both on PC and ARM platform. An algorithm of automatic edge detection was developed to improve the detection accuracy based on Hough Transform. We also proposed a colony automatic counting algorithm based on an improved iterative erosion algorithm to avoid the redundant calculations. The average deviation of counting results was controlled in 6% comparing with gross inspection results.In a word, some significant research and attempt were proposed in the study of colony counting algorithm. A kind of embedded automatic colony counter prototype was developed and the intelligent colony counting algorithm was successfully transplanted on the target embedded platform. Therefore, all of these works not only have theoretical significance but also have extensive practical value for instrument development application.更多还原