| 基于计算机视觉系统的番茄特征提取 |
| |
| 引用本文： | 肖英奎,江力,李因武,栾祥宇,王傲雪.基于计算机视觉系统的番茄特征提取[J].农机化研究,2022,44(2):136-142,148. |
| |
| 作者姓名： | 肖英奎 江力 李因武 栾祥宇 王傲雪 |
| |
| 作者单位： | 吉林大学 生物与农业工程学院/工程仿生教育部重点实验室, 长春 130022 |
| |
| 基金项目： | 吉林省科技厅自然基金项目(20190201019JC)。 |
| |
| 摘 要： | 针对实际番茄特征提取环境复杂情况的问题,提出了针对不同环境应用不同颜色模型来进行阈值分割的方法。通过应用改进的n R-G、YUV两种颜色模型对不同实验环境采集的图像进行阈值分割,并结合canny边缘提取算法、fitzgibbon椭圆拟合算法提取得出番茄像素坐标与像素尺寸,以此完成番茄特征提取。为得出各种颜色模型适用环境等特点,对比各种颜色模型在光线充足果实未被遮挡、光线充足果实部分遮挡和光线较弱果实未被遮挡3种情况下特征提取成功率,并比较3种颜色模型在光线充足果实未被遮挡情况下对采集图像的降噪能力。实验结果表明:n R-G颜色模型适用于采集图像噪声较小的实验环境,对于光线较弱的实验环境该模型表现出较高且稳定的特征提取成功率;YUV颜色模型表现出对含噪图像具有较为稳定的降噪能力,且对光线较强的实验环境表现出较高的特征提取成功率。
|
| 关 键 词： | 番茄特征提取 图像处理 颜色模型 图像噪声 椭圆拟合 |
Tomato Feature Extraction Based on Computer Vision System |
| |
| Xiao Yingkui,Jiang Li,Li Yinwu,Luan Xiangyu,Wang Aoxue.Tomato Feature Extraction Based on Computer Vision System[J].Journal of Agricultural Mechanization Research,2022,44(2):136-142,148. |
| |
| Authors: | Xiao Yingkui Jiang Li Li Yinwu Luan Xiangyu Wang Aoxue |
| |
| Institution: | (School of Biological and Agricultural Engineering,Jilin University/Key Laboratory of Bionic Engineering of Ministry of Education,Changchun 130022,China) |
| |
| Abstract: | In view of the complex environment of actual tomato feature extraction environment,a method of threshold segmentation using different color models for different environments is proposed.Through the application of improved color models of nR-G and YUV,threshold segmentation of images collected in different experimental environments,combined with the canny edge extraction algorithm and the fitzgibbon ellipse fitting algorithm,the tomato pixel coordinates and pixel size are extracted to This completes tomato feature extraction;in order to obtain the characteristics of various color models suitable for the environment,compare the various color models in three cases:full-light fruits are not blocked,full-light fruits are partially blocked,and weak lights are not blocked.The success rate is compared with the noise reduction capacity of the three color models on the collected image when the fruit is not blocked by the light.The experimental results show that the nR-G color model is suitable for the experimental environment with less image noise,and the model shows a higher and stable feature extraction success rate for the experimental environment with weak light.The YUV color model shows a relatively stable noise reduction ability for noisy images,and at the same time shows a high feature extraction success rate for the experimental environment with strong light. |
| |
| Keywords: | tomato feature extraction image processing color model image noise ellipse fitting |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
