框架变换耦合SUSAN的多聚焦图像融合算法

Multi-focus image fusion algorithm based on frame transform coupled SUSAN

为解决当前多聚焦图像融合技术中存在块效应、对比度较低等不足,提出基于框架变换耦合SUSAN(smallest univalue segment assimilating nucleus)图像融合算法。将所有输入图像分解为低频子带和高频子带,利用最小核值相似区SUSAN构建特征提取函数,得到低频和高频特征系数;基于人类视觉系统(human visual system,HVS)特性,分别建立低频和高频子带的融合准则,利用低频特征作为可见性融合措施,高频特征作为融合的HVS模型的纹理信息;利用逆框架变换合成融合图像。实验结果表明,与当前常用多聚焦图像融合技术比较,所提方法获得的融合图像在信息熵、均方根交叉熵、边缘信息评价因子和峰值信噪比的评价指标具有更大优势,融合图像细节完好,视觉质量更高。

For the side effects of the current multi-focus image fusion technology, such as block effect, reduction of contrast and so on,a method of SUSAN coupled frame transform image fusion was proposed. All the input images were decomposed into low frequency and high frequency bands. SUSAN was used to construct feature extraction function, and the characteristic coeffi-cients of low frequency and high frequency were obtained. Based on human visual system characteristics, the fusion criterion of low frequency and high frequency sub bands were established respectively. Low frequency sub band was used as visibility fusion measure,and high frequency sub band was used as the fusion of the texture information of HVS model. The fused image was synthesized through the inverse frame transform. The results show that compared with the current commonly used multi-focus image fusion algorithm, the proposed algorithm can better keep the image details and improve the contrast of the image, which can be used to improve the image visual effects.