基于图像块多向梯度模的小波图像融合方法  

Method for wavelet image fusion based on multi-orientation gradient mode of image block

作　　者：卢振坤[1] 玉振明[2] 

机构地区：[1]梧州学院数理系,广西梧州543002 [2]梧州学院科研处,广西梧州543002

出　　处：《计算机应用研究》2009年第10期3986-3988,共3页Application Research of Computers

基　　金：广西自然科学基金资助项目(0832283);梧州学院基金资助项目(2008D011)

摘　　要：利用块分割图像融合算法的思想,提出了一种基于多向梯度模的小波图像融合方法:对图像块进行小波分解,利用小波反变换得到相同尺度的高频图像块,然后对高频图像块多向梯度滤波,比较对应块的高频图像的总的梯度模值,以挑选出其中清晰的图像块并进行一致性验证,重构图像。该方法不用进行小波反变换重构图像,真正保留了原始的清晰图像。通过实验,证明了该方法对于多聚焦图像的融合具有很好的效果和很好的稳定性。This paper proposed a method for wavelet image fusion based on muhi-orientation gradient mode, by using block partition of the image fusion method of thinking. It adopted wavelet decomposition method to analyze image blocks. High-fre- quency image block with the same scale was got by wavelet inverse transform, and was filtered by the multi-gradient operator. Compared with the total value of the gradient of high-frequency image block was in order to pick out a clear image of the block and to verify the consistency, finally reconstructed the image. This method didn＇ t need wavelet inverse transform to recon- struct image, reserved the original and clear images. Experiments show that the method for muhi-focus image fusion has good results and good stability.

关 键 词：小波变换 图像融合 图像块 多向梯度模 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]