| 基于Gabor滤波器和SVM结合的中华蜂图像识别方法 |
| |
| 引用本文： | 刘超,林寿英,王彩霞,李新亮,张晓青,谢奕涛.基于Gabor滤波器和SVM结合的中华蜂图像识别方法[J].农业工程,2021,11(8):49-52. |
| |
| 作者姓名： | 刘超 林寿英 王彩霞 李新亮 张晓青 谢奕涛 |
| |
| 作者单位： | 1.福建农林大学，福建 福州350002 |
| |
| 摘 要： | 为了减少由于光照强度不一、对比度不同的干扰因素对中华蜂图像识别过程带来的困难，提出一种基于Gabor滤波、PCA降维与SVM相结合的蜜蜂图像识别方法。将人工采集的中华蜂图像进行灰度化、归一化处理，并采用Gabor滤波技术对处理后的图片进行图像特征的提取，进一步通过PCA将高维特征向量进行线性降维，最后将图片特征值矩阵分别经过不同核函数的SVM进行分类识别。通过不同核函数的SVM进行对比建模，测试并分析其对于特征提取后中华蜂图像的建模时长、识别准确率及识别图像时长。试验表明，中华蜂的图像经过Gabor特征提取、PCA降维得到的特征矩阵，经过核函数为Sigmoid的SVM时，其识别特性最好。
|
| 关 键 词： | 中华蜂图像分类 Gabor特征提取 PCA降维 核函数 支持向量机 |
| 收稿时间： | 2021/4/16 0:00:00 |
| 修稿时间： | 2021/5/25 0:00:00 |
Chinese Bee Image Recognition Method Based on Gabor Filter and SVM |
| |
| Abstract: | In order to reduce difficulty of Chinese bee image recognition process caused by interference factors of different light intensities and different contrasts，a bee image recognition method based on combination of Gabor filtering，PCA dimensionality reduction and SVM was proposed.The research was to perform grayscale and normalization processing on artificially collected Chinese bee images，and use Gabor filter technology to extract image features of processed pictures，and further use PCA to linearly reduce high-dimensional feature vectors，and finally the pictures.The matrix was classified and identified through SVM with different kernel functions.The SVM with different kernel functions was used for comparative modeling，and modeling time，recognition accuracy，and image recognition time of the Chinese bee image after feature extraction were tested and analyzed.This experiment showed that the image of Chinese bee was extracted by Gabor feature and reduced by PCA.The feature matrix obtained through SVM with the kernel function of Sigmoid has the best recognition characteristics. |
| |
| Keywords: | |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《农业工程》浏览原始摘要信息 |
| 点击此处可从《农业工程》下载免费的PDF全文 |
