基于改进神经网络的图像融合技术

Image Fusion Technology Based on Improved Neural Network

传统图像融合技术存在融合后图像边缘清晰度低的问题,为此,引入改进神经网络,提出一种图像融合技术的设计方法。通过分析改进神经网络的理论概述,提出基于改进神经网络构建脉冲融合数学模型的方式,对图像融合中的行为进行数学描述与约束。同时,根据每个神经元在网络中的不同状态,制定图像融合行为实施规则,对其附近区域内的神经元信息进行捕获即可。以此根据随机分布概率,对图像中噪声的分布进行定义,在完成边缘去除处理的基础上,实现对图像融合技术的研究。设计对比实验,证明提出的融合技术在对多组图像融合组进行融合后,图像边缘清晰程度明显高于传统融合技术,图像并未出现失真现象。

The traditional image fusion technology has the problem of low definition of image edge after fusion.Therefore,the improved neural network was introduced and the design method of image fusion technology was proposed.By analyzing the theoretical overview of the improved neural network,a mathematical model of pulse fusion based on the improved neural network was proposed to describe and constrain the behavior in image fusion.At the same time,according to the different states of each neuron in the network,the implementation rules of image fusion behavior were formulated,and the information of neurons in the nearby area can be captured.According to the probability of random distribution,the distribution of noise in the image was defined.On the basis of edge removal,the research of image fusion technology was realized.In addition,comparative experiments were designed to prove that the proposed fusion technology has better edge clarity than the traditional fusion technology after fusing multiple groups of images,and the image did not appear distortion.