基于四元数小波变换和自适应神经网络的图像融合处理

Fusion Algorithm of Image Enhancement Based on Quaternion Wavelet Transform and Adaptive Neural Network

夜晚环境下,同一物体的可见光图像、红外图像以及环境噪声等特征信息往往融合在一起,从而导致所提取的图像清晰度变差,边缘模糊等。为改善夜间图像的可视化效果,提出了一种基于四元数小波变换和自适应神经网络的红外与可见光图像融合算法。该算法首先对夜间原图像中的弱光信息进行动态压缩和自适应增强;再利用四元数小波变换对增强后图像的可见光、红外部分进行分解;在低频部分,取系数最大的融合算法;在高频部分,采用神经网络参数自适应融合算法;最后,采用四元数小波逆变换得到融合图像。理论与实验表明,该算法能很好地改善夜视场景下同一目标图像的提取效果。

The visible light image,infrared image and ambient noise from the same object are often fused together in night environment,which results in the deterioration of image definition and blurred edges.In order to improve the visual effect of the night images,an infrared and visible image fusion algorithm based on quaternion wavelet transform and adaptive neural network is proposed.Firstly,the algorithm dynamically compresses and adaptively enhances the weak light information in the night original image;secondly,the visible and infrared parts of the enhanced image are decomposed by using quaternion wavelet transform,the fusion algorithm with the largest coefficient is adopted in low frequency part and the parameter adaptive fusion algorithm of neural network is used in high frequency part;finally,the enhanced image is obtained by using the inverse transform of the quaternion wavelet.The theoretical and experimental results show that the algorithm can improve the extraction effect of the same target image in night condition.