摘要
针对可见光图像与近红外图像特点,提出了一种基于提升小波和分形的多源图像融合方法。首先将已配准的多源图像分别进行提升小波分解,在各层的低频部分用分形维加权平均融合,高频部分用区域交叉信息熵和能量特性融合;再通过提升小波重构得到融合图像。利用苹果树可见光图像和近红外图像进行了实验,实验结果表明,融合后的图像符合视觉特性,综合性能优于传统小波变换融合方法,有利于对图像作进一步分析、理解和识别。
The same object visual and near infrared images were fused in some agriculture pick machine vision systems. A novel fast image fusion algorithm has been proposed based on lifting wavelet transform and fractal dimension theory. Firstly, the registered original images were decomposed by using lifting wavelet transform respectively. Then, the decomposition low frequency components were combined with fractal dimension. The decomposition high frequency components were merged by region cross-entropy and energy features. Finally, the composite image was obtained by using inverse lifting wavelet transform. Experimental results demonstrated that the fusion algorithm is more effective in the fused image quality than traditional method based on wavelet transform. The fused image is suitable to human vision characteristic and is advantageous for further analyzing, understanding and recognizing.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2007年第10期91-93,121,共4页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金资助项目(项目编号:60575020)
湖北省重点学科黄石理工学院机械电子工程学科建设资助项目
关键词
农业采摘机器人
图像融合
提升小波变换
分形维
信息熵
Agriculture pick robot, Image fusion, Lifting wavelet transform, Fractal dimension, Entropy
