基于小波概率估计的图像融合方法研究

在研究了已有的图像融合方法后,提出基于小波变换和最大似然概率估计(MLE)相综合的融合方法,利用概率估计融合模型,首先对不同的传感器图像进行小波分解,然后对相应的子带求解仿射变换参数,根据Bayes规则进行最大后验概率似然估计,得到估计子带系数,最后通过小波反变换得到融合图像.在仿射变换的假设条件下定义融合规则,更适合传感器图像具有局部相反对比度的情况,采用此方法对航空可见光图像和红外图像进行融合实验,其结果与采用其它方法进行了对比,表明该方法的有效性.

Image Fusion on Wavelet Probability Statistic

A probabilistic method for fusion based on wavelet is presented. A Bayesian framework provides for maximum likelihood or maximum a posterior estimates of the true scene from the sensor images .This estimate constitutes the rule for fusing the sensor images. Utilizing wavelet decomposition images into multiple levels, at which approximate the map between the true scene and the sensors by a local affine transformation is defined.A Maximum Likelihood Estimates (MLE) for the parameters of the local affine transformations is given. The merging rule is derived under the assumption of local affine transformation. The fusion results of visible image and infrared image obtained on aerophotography can be performed under local polarity reversals perfectly by our scheme. Experiments and performance show that the method is valid .Abstract A probabilistic method for fusion based on wavelet is presented. A Bayesian framework provides for maximum likelihood or maximum a posterior estimates of the true scene from the sensor images .This estimate constitutes the rule for fusing the sensor images. Utilizing wavelet decomposition images into multiple levels, at which approximate the map between the true scene and the sensors by a local affine transformation is defined.A Maximum Likelihood Estimates (MLE) for the parameters of the local affine transformations is given. The merging rule is derived under the assumption of local affine transformation. The fusion results of visible image and infrared image obtained on aerophotography can be performed under local polarity reversals perfectly by our scheme. Experiments and performance show that the method is valid .