基于小波包变换的图像融合技术的应用

阐述了基于小波包变换的图像融合算法，并对基于像素的融合规则与基于区域的局部方差融合规则作了论述．在此基础上，发展了一类修正后的局部方差对比的融合规则．通过对一混凝土梁构件内模拟的缺陷进行红外探测并进行图像融合，结果表明：对于探测梁内部缺陷，采用小波包变换的融合算法图像不仅充分结合了融合前图像的信息，边缘信息也更为细腻．另外，小波包分解层数对图像融合效果有着重要的影响．通过对加噪声红外图像的融合算例表明：随着小波包分解层数的递增，融合质量却逐渐下降．当分解层数在1～2层时，融合效果比较理想．

Application of Image Fusion Technology Based on the Wavelet Packet Transform

This paper primarily presents image fusion algorithms based on the wavelet packet transform. The two kinds of fusion rules, including pixel.-based and region-based local variance, are discussed, and then modified fusion rules are proposed. An experiment of detecting the simulated damage of a concrete beam by the infrared thermography is carried out. The experimental results show that the fused image based on the wavelet packet transform algorithm not only fully integrates the complementary information from original images, but also demonstrates more elaborate edge information, which can improve qualitative and quantitative evaluation of the image in detecting the interior damage of the concrete beam. In addition, the research indicates that the wavelet package decomposition level is of significance to fusion quality. According to computational examples of infrared images fusion with noises fusion quality decreases with the increase of decomposition levels. This paper suggests when wavelet package decomposition level is equal to 1 or 2, fusion result is more desirable.