【作者】 孙亮；

【导师】 李纪扣；

【作者基本信息】 天津科技大学 ， 检测技术与自动化装置， 2008， 硕士

【摘要】 随着人民生活水平的逐步提高,人们对食品质量的要求日益严格。农产品的加工质量、食用品质和商品价值也在农产品生产过程中越来越突现显著的地位,谷物分选机在我国谷物加工业中的应用也越来越普及。传统的谷物分选机采用硅光电池作为光电探测器,只能通过光线的明暗对谷物颗粒进行识别,局限性很大。为了对谷物分选机进行改进,本文提出一种利用数字图像处理技术和模式识别技术对谷物图像进行处理与识别的方法,利用这种方法替代硅电池的光电流判别法,可以扩充谷物分选机的应用范围,提高分选质量,提高谷物分选机的应用适用性。本论文中以VC++为平台设计数字图像处理系统,运用数字图像处理的方法对采集到的谷物图像进行彩色图像灰度化、中值滤波降噪、灰度拉伸、阈值分割求取二值图像、连通区域标记去除伪区域、八邻域法轮廓提取、最小外接矩形等图像处理操作。通过对谷物图像的数字图像处理操作,成功地提取了谷物图像的灰度、面积、长径、短径等特征值,并计算出谷物的长宽比特征值。针对提取出来的谷物特征值,运用统计模式识别的方法,对谷物特征值进行分析,并求出特征值的阈值区间。针对不同的谷物,可以选取适合的特征值,通过判断该特征值是否在阈值区间内来实现对谷物的识别,达到去除谷物中杂质和异物的目的。在本文的最后,探讨了此方法应用到实际生产中可能会遇到的问题,并对可能会遇到的一些问题提出了处理方法。另外,指出了本论文中的不足之处以及今后继续研究改进的方向。更多还原

【Abstract】 Along with the increase of people’s life level, the request of people to the foodstuffs quality is stricter. The machining quality of farm product, the edible character, and the merchandise value are more and more distinct in the produce process of the farm product. The application of corn selection machine in our machining industry of cereal is more and more popularization. As the traditional machine adapts silicon cell as its photo electricity detector, it distinguishes the cereal only by the brightness of light, it has significant limitations. In order to improve the cereal separator, this paper presents a method that use digital image processing technology and pattern recognition technology to cereal image processing and recognition. Using it instead of optical current silicon cell discrimination, the application of the cereal separator can be expanded and improve the quality of separation, increase the applicability of cereal separator.In this paper, use a VC++ as platform design for digital image processing system, use digital image processing methods to the collection of cereal image to gray the color image, filtering noise, gray stretch, the threshold segmentation strike Binary Image, connectivity regional marker removed pseudo-regional, eight neighborhood contour extraction method, such as the smallest external rectangular image processing. Through the image of the cereal digital image processing operations, we can successfully extract eigenvalues such as cereal image gray area, surface area, long tracks, and short tracks, and calculate the eigenvalues of cereal aspect ratio. Against the characteristic values extracted from the cereal, we use the statistical pattern recognition method, to analyze the cereal eigenvalues, and obtained the threshold interval of the eigenvalues. For different cereals, we can choose the suitable characteristics of value to judge whether they range in the eigenvalues threshold to achieve the recognition of cereals, and to remove impurities of the cereal and foreign body.In the final of this paper, it analyzes the problems that the application of this method to the actual production may encounter, and put forward a few of methods to deal with the problems encountered. In addition, the paper points out the inadequacy of itself and the future improving direction to study.更多还原