运用计算机视觉识别技术进行馒头气孔结构分析  

Analysis of Cell Structure of Steamed Bread by Digital Image Analysis

作　　者：何胜美[1,2] 陈东升[1,3] 张艳[1] 何中虎[1,4] 

机构地区：[1]中国农业科学院作物科学研究所/国家小麦改良中心/国家农作物基因资源与基因改良重大科学工程 [2]广东金融学院基础部,广州510520 [3]宁夏农林科学院农作物研究所,永宁750105 [4]MMYT中国办事处,北京100081

出　　处：《中国农业科学》2007年第1期212-216,共5页Scientia Agricultura Sinica

基　　金：国家"948"重大农业国际合作项目(2003Q01);"973"国家重点基础研究发展规划(2002CB111300)

摘　　要：【目的】利用计算机数字图像分析提取馒头结构的气孔特征,以评价计算机视觉分析在馒头品质评价中的作用。【方法】试验1选用强筋品种Weaver、中筋品种宁春4号和弱筋品种京411,按粉质仪吸水率采用3个加水量处理,共计9个不同样本。图像分析中,采用K-均值算法将气孔从背景中分割出来,提取了3个气孔特征,即气孔总面积、气孔平均面积和气孔总数目,用于数据分析。试验2利用图像分析对11个样本进行馒头品质评价,并与实验室人工主观评分进行比较。【结果】试验1结果表明所取的3个特征能够较好反映加水量和筋力强弱对馒头气孔结构的影响,随着加水量增加和面筋强度增强,气孔总面积增加,这与馒头体积增大一致。试验2中馒头气孔图像特征的评价与人工评价具有较高的一致性,表明计算机图像分析能够较好反映馒头内部结构优劣。【结论】利用图像分析进行馒头品质评价是可行的。[Objective] Cell parameters of steamed bread crumb were obtained by digital image analysis （DIA） with the objective to evaluate the possibility of using computerized vision for steamed bread quality evaluation. [Method] Three wheat cultivars, Weaver with strong gluten, Ningchun 4 with medium gluten, and Jing 411 with weak gluten were used in experiment 1. Three water addition levels based on farinograph＇s water absorption were used, thus 9 samples were included in this experirflent. K-means algorithm was used to segment the steamed bread image. The total cell area, mean cell area, and the number of cells were obtained as characters of steamed bread image. In the second experiment, eleven samples were used to compare the results of machine vision with panel evaluation. [ Result] Results show that those characters are able to characterize the effect of water addition and gluten strength on steamed bread quality. That is, with increase of water addition or gluten strength, the cell total area increases. It corresponds with the change of steamed bread volume. The results of the second experiment indicated that evaluating steamed bread quality with a computer is consistent with manpower evaluation. [ Conclusion] The results show that it is feasible to evaluate steamed bread quality by computerized vision.

关 键 词：图像处理 K-均值算法 馒头品质 加水量 面筋强度 

分 类 号：TS213.2[轻工技术与工程—粮食、油脂及植物蛋白工程]