摘要
采用参数简化的脉冲耦合神经网络(PCNN)分割图像,进行了蝗虫图像分割实验,区域正确识别率达94%,为蝗虫自动侦测系统中的数据处理提供了技术支持。计算机仿真表明,采用PCNN图像分割算法,图像中的目标(蝗虫)易于发现,分割效果明显好于采用开操作处理的图像。
Image binary segmentation is the most fundamental and important preprocessing in image analysis and pattern recognition, which directly affects analyses and results of post-processing. The crucial step in image data processing of automatic locust detection system （ALDS） is image segmentation. A parametrically simplified pulse-coupled neural network （PCNN） was brought forward. Experiments were done on locust images. Area recognition rate （ARR） achieved 94%. The results of computer simulation showed that the objects （locusts） in the image were easier to be found by using PCNN than by the ‘open’ operation. The performance of PCNN in image processing has been tested, and a new approach to detect locusts has been developed.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2007年第10期84-86,107,共4页
Transactions of the Chinese Society for Agricultural Machinery
基金
中国博士后科学基金资助项目(项目编号:20060390545)
国家科技部科研院所社会公益研究专项项目(项目编号:2004DIB3J076)
关键词
蝗虫
图像分割
脉冲耦合神经网络
Locust, Image segmentation, Pulse-coupled neural network
