R/G/B直方图对比算法判别红茶发酵适度  

Black tea fermentation degree monitoring with histogram comparison algorithm

作　　者：李莎莎[1] 洪文娟 陈华才[1] 董春旺[2] 叶阳[2] 

机构地区：[1]中国计量大学光学与电子科技学院,浙江杭州310018 [2]中国农业科学院茶叶研究所,浙江杭州310008

出　　处：《中国计量学院学报》2016年第2期172-176,共5页Journal of China Jiliang University

基　　金：国家科技支撑计划项目(No.2012BAF07B05);浙江省重点科技创新团队项目(No.2010R50028)

摘　　要：提出了一种基于R/G/B直方图对比算法判别红茶发酵适度的方法.设计了一套采集红茶发酵图像采集系统采集红茶发酵叶的颜色图像.通过计算红茶发酵叶在制品图像颜色分量直方图与标准直方图的匹配度判别发酵适度,设定发酵适度的标准图像与阈值T,并实时采集分析红茶发酵过程中在制品图像的R/G/B三种颜色分量直方图,计算其与标准图像的Manhattan距离.当Manhattan距离D的均值小于设定阈值T时,则判别为发酵适度,并输出信号停止红茶继续发酵.研究结果表明,基于R/G/B直方图对比算法判别红茶发酵适度状态是可行的,与感官评审结果对比,其判别准确率达到93.2%.We proposed a method to discriminate black tea fermentation degree with the image color component R/G/B histogram. An image acquisition system was established to collect the images of the fermenting black tea. The standard image of moderate fermentation and the threshold value of T were preset. Then the R/G/B component histograms of black tea images in the fermentation process were collected. The Manhattan distances of the images from the standard image were calculated. When the average Manhattan distance D was less than the threshold value T, the fermentation was moderate. The results show that it is feasible to discriminate the black tea fermentation moderate condition base on the R/G/B histogram comparison algorithm. The overall discrimination accuracy rate is 93. 2% compared with the results of sensory evaluation.

关 键 词：红茶 发酵适度 Manhattan距离 图像处理 感官评价 

分 类 号：TS272[农业科学—茶叶生产加工]