预处理器
人工智能
计算机科学
过程(计算)
模式识别(心理学)
计算机视觉
特征提取
操作系统
作者
Yana Tang,Shilong Zhang
出处
期刊:Journal of Testing and Evaluation
[ASTM International]
日期:2022-06-21
卷期号:51 (3): 1277-1294
摘要
Abstract In order to solve the problems of traditional recognition and monitoring methods, such as low recognition accuracy and long monitoring process, in this paper, a real-time monitoring method of vegetable diseases and pests based on optical fiber sensor is proposed. Firstly, according to the optical fiber sensor imaging technology, the crop image to be detected is obtained, and the preprocessing steps such as feature extraction are carried out at the same time to obtain three single classifiers composed of image color characteristics, texture macro description, and detail expression; secondly, the parallel pattern is used to fuse the three classifiers, and then a crop pest recognition method combined with multiple classifiers is proposed according to the K-nearest neighbor method; finally, based on the recognition results, the video monitoring dynamics of pests is constructed. In the construction process, the open design is implemented combined with the optical fiber sensor, and the corresponding pest data table is constructed according to the required data to realize the real-time monitoring of vegetable pests and diseases. The simulation results show that this method can effectively solve the problems faced by traditional methods in the experiment and has higher advantages compared with other methods in the calculation process.
科研通智能强力驱动
Strongly Powered by AbleSci AI