【作者】 孙宗保；

【导师】 姜松； 蔡健荣；

【作者基本信息】 江苏大学 ， 农产品加工及贮藏工程， 2003， 硕士

【摘要】 粒度是粉体物料的重要质量指标，对产品的许多技术性能和实际应用具有重要的指导意义。目前对粉体粒度的测量大多采用沉降法等常规方法，存在测量时间长、测量步骤多、测量准确度受主观因素影响大等许多不利因素。基于计算机视觉的粒度检测技术可以有效克服常规方法的不足，获得科学的检测结果，代表了粒度检测的发展方向。 本文根据粒度图像的特点，采用非线性对比度增强处理改善颗粒显微图像的对比度；对多种阈值分割方法进行比较，分析了它们的性能和用于粒度图像的适用性，在此基础上提出了一种基于梯度图像的动态阈值分割方法；对获取的颗粒二值图像进行形态滤波和空洞填充，设计了识别算法去除凝聚体；对重叠不太严重的粘结颗粒采用了两种基于数学形态学的分割算法；提取了周长、面积、粒径、复杂度、最长Feret径、最短Feret径等大小和形状特征参数，对珍珠粉等粉体粒度进行了测量。 实验表明用显微图像进行粒度检测，测定分布状态直观明了，检测速度快、结果稳定可靠，可广泛应用于0.1～150μm的各种物料的颗粒粒径的检测。更多还原

【Abstract】 Particle size is an important quality index of powdery material. It exerts significant influence on technical performance of products and plays key role in practical application. At present the routine methods, for example, sedimentation and fractionation technique are chiefly adopted in measurement of particle size. There are some disadvantages in these routine methods, for example, they takes many steps and costs much time, and the accuracy of results is effected greatly by subjective factors of human inspectors. The measurement technology of particle size’s using computer vision can overcome the shortage of routine methods, and so it is one of the most effective methods for measurement of particle size.In this paper, the contrast of original image is enhanced by nonlinear transformation based on the characteristics of grain image; After comparing several methods of threshold segmentation and analyzing their performance and applicability on grain image, the adaptive threshold segmentation based on gradient image is presented; The noise of binary image is reduced with morphological filter, the hollow in binary image is filled in and a recognition algorithm is designed to remove agglomerate particle; on the basis of mathematical morphology two methods are adopted to segment agglomerate grains that are not agglomerated tightly; The parameters of size and shape, perimeter, area, diameter, complexlity, longest feret diameter, shortest feret diameter etc, are extracted. This technique is applied in measurement of pearl powder’s particle size.Experiments proved that this method of micro-image inspection of particle size could determine grain distribution clearly and directly. The device can detect particle sizewidely ranged from 0.1 to 150μm and the result of inspection is speedy and stable.更多还原