摘要
深加工是目前我国水果产业发展的主方向。设计了基于神经网络的自动水果分类系统,包括系统的硬件构成和分类程序。其中图像的特征提取是水果分类程序的前提。提取了水果图像的形状、颜色、纹理3种主要特征,将3种特征向量输入到神经网络分类器进行分类识别。仿真结果表明,该方法实时性好,分类准确率高,可满足水果深加工生产的实用需要。
Fruit＇ s deep processing is currently the main direction of the fruit industrial development in China. An automatic classification system of fruit based on neural network was designed, including the hardware composition and classification program of the system. Feature extraction is the premise of the classification program of fruits. Three main features such as figure, color and texture of the fruit images were extracted. They were input to BP neural network classifier for classification. The simulation results showed that this technique was of good real-time and high rate of classified accuracy and it could meet the practical needs of fruit＇ s deep processing.
出处
《安徽农业科学》
CAS
北大核心
2009年第35期17392-17394,17439,共4页
Journal of Anhui Agricultural Sciences
基金
广东省自然科学基金项目(8152902001000014)
广东省高等学校自然科学重点研究项目(05z025)
关键词
特征提取
特征向量
BP神经网路
分类
Feature extraction
Feature vector
BP neural network
Classification
作者简介
吕秋霞（1978-），女，湖北公安人，硕士，讲师，从事智能信息处理、视频检测研究。
