基于支持向量机的红外与可见光图像融合算法

Infrared and Visible Image Fusion based on Support Vector Machine

针对红外和可见光图像具有的不同特点,提出一种基于支持向量机的红外与可见光图像融合新算法,首先,利用支持向量多尺度变换对红外图像和可见光图像进行分解,得到不同分辨率下的高频图像和低频图像,然后对低频图像采用区域能量与匹配度加权相结合的方法,利用区域像素相关性,根据匹配度选择权重值进行融合,充分保留图像轮廓特征信息;对高频图像采用多方向拉普拉斯算子计算邻域加权和,结合局部能量的融合策略,尽可能突出纹理细节信息;最后对融合得到的高频和低频系数进行重构得到融合图像。实验结果表明,相对于传统红外与可见光图像融合方法,该算法有效综合了源图像的纹理结构信息,并在主观视觉与客观指标上获得了更好的融合效果。

According to the different characteristics of infrared and visible images,a new infrared and visible image fusion algorithm based on support vector machine is proposed.Firstly,support vector transform is used to decompose the infrared and visible images at multiple scales to obtain the high-frequency and low-frequency coefficients of the images at different scales,and then the regional energy is combined with the matching degree weighting method for the low-frequency coefficients,Using the correlation of regional pixels,the weight value is selected according to the matching degree to fuse,and the feature information of image contour is fully preserved;For high-frequency coefficients,multi-directional Laplace operator is used to calculate the weighted sum of neighborhood,and the fusion strategy combined with local energy is used to highlight the details as much as possible;Finally,the fused image is obtained by reconstructing the fused high-frequency and low-frequency coefficients.The experimental results show that compared with the traditional infrared and visible image fusion methods,this algorithm effectively integrates the texture structure information of the source image,and achieves better fusion results in subjective and objective indicators.