摘要
针对传统的离散小波变换融合算法在图像边缘特征表达上存在不足,所以采用具有平移不变性,多方向选择性等具有显著性特点的双树复小波变换对多聚焦图像进行分解,按一定规则融合后,再重构。在融合规则上,以研究高频分量为重点。低频分量只采用较为简单的加权平均法;高频则采用以标准差作为融合测度,分别提出了两种图像融合算法。通过在性能参数上,对两种实验结果进行评价和对比,最终得到图像融合的最佳算法。
For the deficiency of conventional discrete wavelet transform fusion algorithm in features of the image edge, DT-CWT some notable features, such as translational invariance, polydirectional selectivity and so on, is adopted to decompose and reconstruct the multi-focus images. According to the fusion rule, the study is focused on the high-frequency component. The low-frequency component only employs simple weighted average method while the high frequency component uses the standard deviation as the fusion measure, and based on this, two kinds of image fusion algorithms are proposed. These algorithms are evaluated and compared in their performance parameters, and finally the optimum image fusion algorithm is achieved
出处
《通信技术》
2011年第12期104-106,共3页
Communications Technology
关键词
边缘特征
平移不变性
融合规则
edge feature
translation invariance
fusion rule
