基于计算机视觉技术的梗丝形态表征方法

Morphology characterization of cut stem based on computer vision technology

为研究梗丝形态表征方法,采用图像分析软件对CCD相机获取的梗丝二维图像进行处理,利用拟合方程计算出梗丝的均匀性系数和特征宽度,结合梗丝宽度分布区间建立了梗丝形态的表征方法。结果表明：1用叶丝宽度分布方程对8个梗丝样品宽度拟合,其决定系数R2均超过0.98,且残差均在±0.1以内,适用于梗丝宽度分布的计算。2采用梗丝特征宽度、占比最高的宽度分布区间2项指标表征梗丝形态,将梗丝形态划分为4种,丝状梗丝特征宽度y≤1.2 mm,占比最高的梗丝宽度区间为0.6～1.2 mm;片状梗丝特征宽度y≥1.8 mm,占比最高的梗丝宽度区间为〉1.8 mm;近丝状梗丝特征宽度1.2，

The 2-Dimensional digital images of cut stems captured by a CCD camera were processed with imageanalysis software,the uniformity coefficient and characteristic width of cut stems were computed by fittingequation,and the method for morphology characterization of cut stems was developed by combining with thedistributing ranges of cut stem width.The results showed that：1） Fitting the width of 8 cut stem samples with thefunction representing cut strip width distribution,the determination coefficients（R2） all exceeded 0.98 withresidual errors within ± 0.1,it indicated that this function also applied to cut stem width.2） By the characteristicwidth and the main width range of cut stem,the morphologies of cut stems were divided into 4 types：filamentouscut stem,its characteristic width y was ≤1.2 mm and the main width range（WR） was 0.6-1.2 mm; flake cutstem,y≥1.8 mm and WR〉1.8 mm; filament-like cut stem,1.2〈y〈1.8 mm and WR 0.9-1.5 mm; flake-like cutstem,1.2〈y〈1.8 mm and WR 1.2-1.8 mm.3） When the uniformity coefficient was ≥5.5,the samples had betteruniformity.