基于MSTO的含噪声多传感器图像融合算法

Fusion algorithm with multi-sensor noisy image based on MSTO

为了解决在含噪声多源传感器图像融合中,常规滤波存在图像边缘缺失、对比度差的缺点,提出了一种基于多尺度顺序开关算子(multi-scale sequential toggle operator,MSTO)和Beamlet保边滤波算子的含噪声红外与可见光图像融合算法.首先,将多源图像通过MSTO进行多尺度分解,得到能量分量和细节分量.对于细节分量采用Beamlet保边滤波算子进行处理,保持图像边缘细节的同时滤除噪声,采用MSTO计算出能量图像的亮边缘和暗边缘并融合叠加到细节分量中,进一步增强融合图像的边缘.对于能量分量采用基于灰度值取大的融合规则.最后根据MSTO反变换对融合后的能量分量和细节分量进行重构,得到结果图像.实验结果表明,融合后的图像不但滤除了噪声,而且对轮廓和边缘细节得到较完整的提取和增强.该图像融合算法在含噪声多源传感器的融合中取得较好的效果.

In the multi-sensor noisy image fusion,it was easy to obtain the fused images with loss of image edges and low image contrast based on the general filter methods.A novel fusion algorithm with noisy infrared and visible light images was proposed based on a multi-scale sequential toggle operator(MSTO) and an improved bilateral filter method.First,the energy component and the detail component were obtained by MSTO multi-scale decomposition.The detail component was processed by Beamlet operator to filter noises while keeping edge information on the images.Then,the bright edge image and dark edge image with the energy image were calculated by MSTO,and added to the detail component to enhance edges.The maximum rule was used in the energy component fusion.MSTO inverse transform was used to decompose the fused detail component and the energy component.The experimental results show that method filters the noise,and extracts and enhances the contour and the edge details.The image fusion algorithm is effective in the multi-sensor noisy image fusion.