基于双树复小波变换的夜视图像融合

Dual-tree Complex Wavelet Transform Based Night Vision Images Fusion

针对夜视图像融合的特点:尽量获得精确的特征定位和清晰的图像表达,提出了基于双树复小波变换的融合方法,充分利用了双树复小波变换所具有的平移不变性、方向选择性、有限的数据冗余、完美的重构性和较高的计算效率等特点。并在此基础上提出了多策略的融合规则:低频采用清晰度,高频采用窗口能量。除了目视定性评价外,还使用了通用的基于主客观的定量评价方法。通过对两组夜视图像融合的实验测试,表明了双树复小波变换相对于离散小波变换的优势和高低频采用不同融合规则的有效性。

An image fusion method based on dual-tree complex wavelet transform was put forward aiming at the characteristic of night vision images fusion which is obtaining exact position and clear image presentation. The dual-tree complex wavelet transform is applied in fusion processing because of it＇s properties which include shift invariance, directional selectivity, limited redundancy, perfect reconstruction and computational efficiency. The fusion rules based on multi-scheme were also presented. The low frequency part employed definition, and the high frequency parts employed window energy. Besides the person＇s vision evaluation, the universal measures which combines objective and subjective factors were also used. The experimental results of two groups night vision images fusion show that the dual-tree complex wavelet transform is superior to discrete wavelet transform and the fusion rules of high and low frequency utilizing different fusion measurement are effective.