基于小波变换的自适应图像融合算法

An adaptive image fusion algorithm based on wavelet transform

针对现有图像融合方法存在的光谱信息和空间细节信息不能较好兼顾的问题,建立了图像内容自适应的融合准则和一致性选取准则,提出了一种基于小波变换的自适应图像融合算法,实现了多光谱图像与全色图像的融合,并对融合图像进行了主、客观评价。着重从图像融合如何提高目标的区分度和识别率的角度给出主观评价,通过光谱扭曲度、清晰度客观分析多光谱与全色图像的融合效果。实验结果表明,该算法充分利用了全色图像的空间细节特征、图像边缘和方向性特征信息,保留了多光谱图像的光谱信息特征,提高了融合图像的主观效果,有利于信息的提取和目标解译。在光谱和空间细节综合保持方面,该算法优于IHS融合方法和传统的二进小波融合方法。

Aiming at present fusion methods, worse tradeoff between spectrum information and spatial details, the paper estab- lishes the image content self-adaptation fusion rules and conformability choice criteria, and realizes the fusion system between multi-spectral and panchromatic remote sensing images. The fusion images are evaluated by subjective and objec- tive methods. The paper subjectively appraises the fusion images on how to improve the discrimination between objects and background, and the recognition efficiency. The fusion effect is objectively analyzed by spectrum resolution and defi- nition. The experimental data show that this algorithm makes the best of spatial detail feature, image edge and directional feature information for panchromatic images, retains spectrum information for multi-spectral images, achieves better visual quality, and it is in favor of information extraction and image interpretation. The experiments prove that this algorithm is better than IHS and traditional dyadic wavelet fusion methods in spectrum maintenance and spatial detail reservation.