摘要
种子成熟度需要受过长期训练的专家通过肉眼进行观察和判断。为了改变传统人工经验判断的方式,该文提出了一种基于Gabor小波特征提取及深度神经网络的葡萄种子图像分类识别算法,以便实现高效、准确的分类识别效果。首先,利用背景差分法在背景图像中分割出兴趣目标,从而完成图像的预处理。然后,通过改进的Gabor小波特征提取,使得Gabor滤波后的图像具有更多的细节纹理信息。最后,将深度卷积神经网络和提取到的纹理特征信息相结合进行分类。实验结果表明,基于机器学习的葡萄种子成熟度识别是切实可行的。且相比于其他类似分类算法,本文算法的图像分类精度有了一定的改善。
Seed maturity has a great influence on the quality of wine, and it needs to be observed and judged by the naked eye by experts who have been trained for a long time. In order to change the way of traditional artificial experience judgment, a grape seed image classification and recognition algorithm based on Gabor wavelet feature extraction and deep neural network is proposed to achieve efficient and accurate classification and recognition. First, the background difference method is used to segment the interest target in the background image,thereby completing the image preprocessing. Then, the improved Gabor wavelet feature extraction makes the Gabor filtered image have more detailed texture information. Finally, the deep convolutional neural network and the extracted texture feature information are combined to classify. The experimental results show that the recognition of grape seed maturity based on machine learning is feasible. The proposed image classification accuracy exhibits a certain improvement compared with other similar classification algorithms.
作者
杨旺功
淮永建
张福泉
YANG Wang-gong;HUAI Yong-jian;ZHANG Fu-quan(School of Information,Beijing Forestry University,Haidian Beijing 100083;School of Computer Science&Technology,Beijing Institute of Technology,Haidian Beijing 100081)
出处
《电子科技大学学报》
EI
CAS
CSCD
北大核心
2020年第1期131-138,共8页
Journal of University of Electronic Science and Technology of China
基金
国家自然科学基金(31770589)
中央高校科研团队建设项目(2015ZCQ-XX)
关键词
分类识别
深度学习
特征提取
GABOR小波
葡萄种子
机器学习
神经网络
classification recognition
deep learning
feature extraction
Gabor wavelet
grape seed
machine learning
neural network