基于计算机视觉核桃质量预测方法的研究

Research on Weight Prediction Method of Walnut Based on Computer Vision

为研究核桃质量在线检测的方法,以"礼品2号"核桃为样本,构建图像采集系统采集样本图像信息,对图像进行预处理、形态逻辑学算法获取样本投影面积,并建立最小二乘支持向量机(LS-SVM)和多项式回归方程质量预测模型。结果表明,LS-SVM质量预测模型相关系数为0.897 4,相对误差均值为6.5%;一元二次多项式回归方程质量预测模型决定系数为0.857 207,相对误差均值为5.9%,均满足在线质量检测要求。基于计算机视觉技术实现对核桃质量的预测,为实现核桃在线分选提供理论基础。

To study the method of on-line detection of walnut weight.This study used“lipin No.2”walnut as a test sample to construct an image acquisition system to collect sample image information,preprocess the image,and obtain a sample projection by morphological logic algorithm area,and establish a least squares support vector machine(LS-SVM)and polynomial regression equation weight prediction model.The results showed that the correlation coefficient of ls-svm weight prediction model was 0.897 4,and the average relative error was 6.5%.The determination coefficient of the weight prediction model of quadratic regression equation with one unknown was 0.857 207,and the mean relative error was 5.9%,which all meat the requirements of online weight detection.Based on computer vision technology,this study can predict the weight of walnut and provide a theoretical basis for online separation of walnut.