摘要
以提高玉米叶部病害检测精度为目标,提出一种基于并行PCNN的玉米病害彩色图像非监督分割方法。该方法是在CIE LUV颜色空间中以归一化的L+U特征值为外部激励输入,以邻域像素间几何距离与色度差的综合信息为PCNN耦合连接域权值,以颜色矢量的最小色差对比度为最佳分割结果判别准则,用改进型并行PCNN对玉米病害彩色图像进行分割。对4种病害100幅图像的分割实验表明,该方法分割效果较好,适应度较高,参数设置复杂度低。
A kind of unsupervised segmentation processing method based on parallelized firing PCNN algorithm was proposed.The color images of corn disease were segmented by improved parallelized firing PCNN which the normalized L＋U as external stimulus input,the integrated information of the geometric distance and the color difference between neighboring pixels as the PCNN coupling value,the minimum color contrast of color vector as the criteria of the best segmentation results,in parallel with improved disease of maize PCNN to segment color images.The segmentation experiments which 100 images of four kinds of diseases showed that the method could better segment the diseased regions with high fitness and low complexity parameters.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2011年第11期148-153,共6页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金资助项目(60975007)
关键词
玉米病害
图像处理
并行PCNN
彩色图像分割
Corn diseases
Image processing
Parallelized firing PCNN
Color image segmentation
