Method of Ear Detection for Maize Seed Based on Fisher Criterion

基于改进的Fisher准则玉米种子果穗检测方法

¹ School of Automation, Northwestern Polytechnical University, Xi'an 710072, China
² School of Mechanical and Electrical Engineering, Xi'an University of Architecture and Technology, Xi'an 710055, China
³ School of Marine Science and Technology, Northwestern Polytechnical University, Xi'an 710072, China

Received: 1 April 2017

Abstract

The purpose of this paper is to use machine vision to detect the size, shape, texture and color of maize seed ear. The traditional method is very limited and cannot meet the detection efficiency. This paper used logistic regression linear discriminant analysis method, the rule of perceptron, Fisher method and the least-square method, this paper use change between the Fisher criterion according to the the class of discrete degree error matrix and matrix of the discrete degree of reconstruction error in class than the maximum projection direction, classifying clusters. By fusion Fisher discriminant analysis method for testing in the test, through a large number of experiments and comparison method, the experiment proved that Fisher can high efficiency and high precision of classifying seed corn ear detection.

摘要

采用机器视觉检测玉米种子果穗的大小、形状、纹理和颜色等特征，传统方法检测非常有限而且不能满足检测效率。采用了logistic回归线性判别分析方法、感知机准则、最小二乘法和Fisher方法对比，利用变化的Fisher准则根据类间离散度误差矩阵与类内离散度重建误差矩阵之比的最大值确定投影方向，对果穗进行分类。通过融合Fisher判别分析方法进行检测中进行检测，通过大量实验和其他方法对比，实验证明Fisher能够高效率高精度的对玉米种子果穗进行分类检测。