【摘要】 图像融合结合图像处理、信号处理、计算机和人工智能等相关技术.通过对多源图像数据信息的提取合成,从而获得同一场景目标较为准确全面的图像描述.神经网络具有强大的非线性映射逼近能力,将神经网络用来进行滤波融合,避免了传统滤波图像变模糊问题.通过小波神经网络自适应动量快速学习算法进行图像滤波融合,能从根本上避免局部最优,且加快收敛速度,具有很强的学习和泛化能力,也避免了网络结构盲目设计.仿真实验表明,用本方法实现的融合图像更加符合人的视觉特性.更多还原

【Abstract】 Image fusion is a combination of image processing,signal processing,computer and artificial intelligence.Through the extraction and synthesis of information from multiple images of the same scene objects,our research obtains the more accurate and comprehensive description of the image.Neural network has the strong approximation capability of nonlinear mapping and is used for filtering and fusion to avoid the blurring problem by traditionally filtering image.The wavelet neural network makes use of self-adaption momentum fast training algorithm for image fusion,which can effectively avoid local optimization and increase the convergence rate.The wavelet neural network has strong learning and generalization ability and avoids the blindness of network design.The experimental simulation shows that the image after fusion is more consistent with human visual characteristics.更多还原