基于多阈值分割技术的玉米角质率定量测定方法研究  

Quantitative Determination of Maize Endosperm Vitreousness Based on Multi-Threshold Segmentation

作　　者：刘双喜[1] 傅生辉[2] 王金星[1] 范连祥 牟华伟 孟凡荣[2] 张春庆[3] 

机构地区：[1]山东农业大学山东省园艺机械与装备重点试验室,泰安271018 [2]山东农业大学机械与电子工程学院,泰安271018 [3]山东农业大学农学院,泰安271018

出　　处：《中国粮油学报》2016年第9期141-145,共5页Journal of the Chinese Cereals and Oils Association

基　　金：国家公益性行业专项计划(201303005)

摘　　要：为解决人工测定玉米角质率工序复杂、精度低的问题,提出一种基于多段阈值分割技术的玉米角质率定量测定方法。通过对玉米籽粒进行纵剖染色处理,将玉米胚部与其他部位进行区分,然后对样本剖面图像进行预处理,并且利用多段阈值分割方法对玉米籽粒进行多次阈值分割,分别得到玉米角质区、粉质区图像,统计各区域像素数,得到各区域面积,计算可求得玉米角质率。试验表明,该方法测试速度优于人工测试,可达到3-10 min/100粒,且变异系数在2.0以下,方法稳定性好,面积误差可精确到每像素0.002 mm2,保证了测试的有效性。该方法操作简便且不受人为因素干扰,可为玉米角质率测定及玉米品质定级提供有效手段。At present, the way to get the maize endosperm vitreousness by manual measurement is too compli- cated and inefficiency to be used widely. In order to solve the problem, the paper puts forward a method for the quan- titative determination of maize endosperm vitreousness based on multi - threshold segmentation. At first, the maize embryo can been distinguished from the maize seed by dying with the reagent. And then the image of single maize seed is gotten after image preprocessing. The different areas of maize seeds image are extracted by multi - threshold segmentation including keratin endosperm part and farinaceous albumen part. Finally, the actual area of every part can be gotten by calculating the number of pixels in the region which can calculate the percentage of maize endosperm vitreousness. The experiment result shows that the method to determine the maize vitreous rate by the machine vision is faster than manual testing, it only takes 3 - 10 rain to test one hundred seeds. And the stability of the method is better which the variance coefficient is below 2.0 and the computational accuracy of the area reaches to 0.002 mm2. This method is simple and reduces man - made factors, and can provide an effective way for the vitreousness determi- nation and grading of maize quality.

关 键 词：玉米角质率 预处理 多段阈值分割 角质 粉质 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]