运用计算机视觉识别技术进行馒头气孔结构分析

【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。

Analysis of Cell Structure of Steamed Bread by Digital Image Analysis

Cell parameters of steamed bread crumb were obtained by digital image analysis (DIA), with the objective to evaluate the possibility of using computer machine vision for steamed bread quality evaluation. Three wheat cultivars, Weaver with strong gluten, Ningchun 4 with medium gluten, and Jing 411 with weak gluten were used. Three water addition levels based on farinograph's water absorption were used, thus 9 samples were included in this experiment. K-means algorithm was used to segmentation of steamed bread image. The total cell area, mean cell area and the number of cell were obtained as characters of steamed bread image. Results show that those characters are able to characterize the effect of water addition and gluten strength on steamed bread quality, i.e., with increase of water addition or gluten strength, the cell total area increases. This is corresponded with the change of steamed bread volume. Eleven samples were used to compare the results of machine vision with panel evaluation, and result indicates that it is feasible to evaluate steamed bread quality by computer machine vision.