一种基于微观组织特征的粉丝品质检测研究

Inspecting qualities of starch-noodle based on microstructure of processed starches

粉丝的食用品质和价格随原料淀粉种类的不同而有显著的差异。目前,商检中主要是依靠感官评价,可靠性差。本文利用扫描电子显微镜得到粉丝组织的显微图像,运用图像处理技术,提取了基于灰度共生矩阵及分形理论的显微图像的微观特征,以神经网络为模式识别分类器,建立了绿豆、玉米及其混合淀粉制成的“龙口粉丝”品质自动检测系统。实验结果表明,方法可行,并取得了较好的结果。

The edible quality and price are very different in various starch-noodles, which mainly depends on the kind of starches from which the starch-noodle is made. At present, the inspection, classification of starch-noodle and the component analysis of starch in starch-noodle mainly rely on sensory perception, which is fallibility or trustless. An automatic inspection system for Longkou starch noodle made from starch of mung been, corn and their mixture was established. The system classifies starch-noodles based on recognizing the microstructure pattern of starch and components in starch-noodle. The proposed method consists of three step： 1） to obtain the micrograph of starch-noodle with scanning electron microscopy. 2） To extract features of fractal geometry and gray-level co-occurrence matrix from the micrograph. 3） To distinguish starch-noodle using these combined features as input vectors of artificial neural network. Experimental results show that the proposed method is feasible and effective.