基于支持向量机和压力传感器的水果分类系统

Fruit classification system based on the pressure sensorcoupled with support vector machine

目的:提高水果分类准确率并兼顾低计算成本和低传感器成本。方法:提出一种基于电容式压力传感器的水果分类系统,通过采用高斯核函数的支持向量机算法(高斯支持向量机算法)对水果进行分类。所采用的电容式压力传感器由薄铜片和一层乙酸乙烯酯制成,压力传感器被固定在模拟机械手的聚酰胺氨纶手套的拇指和食指上。获得的电容值以数字电平的形式表示,通过数据处理软件进行数据提取,采用核函数的支持向量机算法对电容数据进行处理,以确定水果的类别。结果:对11种水果的分类结果表明,采用高斯—支持向量机算法的智能手套能够实现水果的高准确率分类,且可根据实际需要在分类准确率、计算成本和低传感器成本之间权衡。结论:该系统可用于开发水果分类的电子系统,提升分类机械手的水果分类性能。

Objective:To improve fruit classification accuracy while considering low computational cost and low sensor cost.Methods:A fruit classification system based on capacitive pressure sensor was proposed.The system used support vector machine algorithm with a Gaussian kernel function(GKF-SVM)to classify fruits.The capacitive pressure sensors used were made of thin copper sheets and a layer of vinyl acetate,and these sensors were fixed to the thumb and index finger of a polyamide spandex glove that simulates a robotic hand.The obtained capacitance value was expressed in the form of digital level,and the data was extracted by data processing software,and the capacitance data was processed by SVM algorithm with kernel functions to determine the category of a fruit.Results:The classification results of 11 kinds of fruits in the designed classification system shown that the smart glove using GKF-SVM algorithm could achieve high accuracy classification of fruits,which could trade off between classification accuracy,calculation cost and low sensor cost according to actual fruit classification needs.Conclusion:The research results can be used to develop electronic systems for fruit classification to improve the fruit classification performance of classification manipulators.