融合多维特征的稻米外观品质检测技术  

Rice Appearance Quality Detection Technology Based on Muti-demensional Features

作　　者：邢键 罗佳顺 XING Jian;LUO Jia-shun(College of Information and Computer Engineering,Northeast Forestry University,Harbin 150040)

机构地区：[1]东北林业大学信息与计算机工程学院,哈尔滨150040

出　　处：《哈尔滨理工大学学报》2021年第5期76-82,共7页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(61405045,31470714,61975028);中央高校基础研究基金(2572017DB04).

摘　　要：图像处理技术用于稻米外观品质的检测具有效率高的优点,但其易受弱光照强度的影响。为了提高图像质量,提出了一种新的数据融合处理算法,实现了大米样品和背景的分割,最大限度地消除了噪声,提高了后续检测函数的精度。实现了一套碎米、裂纹、垩白度和加工精度的计算机自动识别功能。本实验选取6种大米作为试验样品,随机抽取10粒大米作为一组,每次试验选取4组样品。经过多次试验,结果表明,该系统对随机稻谷样品的碎米率检测准确率为97.01%、稻种检测准确率为97.60%、裂纹检测准确率为98.22%,优于传统人工检测方法。该系统为进一步完善稻米品质自动检测技术提供了技术依据。Image processing technology has the advantage of high efficiency in rice appearance quality detection,but it is easily affected by weak light intensity.In order to improve the image quality,a new data fusion algorithm is proposed,which realizes the segmentation of rice sample and background,eliminates the noise to the maximum extent,and improves the accuracy of subsequent detection function.A set of automatic recognition functions of broken rice,crack,chalkiness and machining accuracy are realized.In this experiment,6 kinds of rice were selected as test samples,10 grains of rice were randomly selected as a group,and 4 groups of samples were selected for each test.After many tests,the results show that the detection accuracy of the system for broken rice rate of random rice samples is 97.01%,rice seed detection accuracy is 97.60%,crack detection accuracy is 98.22%,which is better than the traditional manual detection method.The system provides a technical basis for further improving the automatic detection technology of rice quality.

关 键 词：图像处理 大米外观 自动检测 数据融合 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] S24[自动化与计算机技术—计算机科学与技术]