图像融合算法的研究与实现.docVIP

PAGE II 摘 要 数字图像融合是将两个或两个以上的传感器所获得的同一场景的多幅图像进行融合；或者是将同一传感器在不同时刻获得的同一场景的多幅图像加以综合，生成一个新的图像的过程。图像融合技术可以有效的减少被感知对象或环境解释中存在的不确定性和误差，最大限度的利用源图像的信息，有利于对图像进行增强、特征提取、去噪、目标识别与跟踪及三维重构。本文主要研究了一种基于小波变换的图像融合方法，首先利用小波变换将源图像分解成高频分量和低频分量，在高频域内，选择邻域平均绝对值较大的多源图像系数作为重要小波系数；在低频域内，新的逼近系数通过对多源图像的小波系数进行加权平均得到，然后对重要小波系数和加权逼近系数进行小波反变换，即可得到融合之后的图像。使用MATLAB软件对本文研究的融合算法进行实现和仿真，实验结果表明通过基于小波变换的图像融合方法得到的融合图像更加清晰，更具有识别力且图像质量得到了较大提高，并可广泛应用于多种研究领域。 关键词：图像融合；小波变换；MATLAB ABSTRACT Digital image fusion is the process, which combines many images from two or more sensors in the same scene or from the same sensors in different time in the same scene. The image fusion technology can effectively reduce the uncertainty and error that is existed in cognitive object or environmental interpretation. The image fusion technology can also make the most of the source image information to make the image enhancement, feature extraction, denoising, target identification, tracking and 3D reconstruction. This paper mainly researches the image fusion method which is based on wavelet transform. Firstly, it uses wavelet transform to break down the source image into low-frequency and high-frequency components. Secondly, in the high frequency domain decomposition, the paper selects the larger coefficient in average absolute value multi-source image as important wavelet coefficients. Thirdly, in the low frequency domain, through the weighted average from the wavelet coefficients of the multi-source image, this paper gets the new approximate coefficient. Finally, using the important wavelet coeffients and weighted approximate coefficient to make the wavelet transform, we get the image after image fusion. Using MATLAB software to implement and simulate the fusion algorithm that this design researched, the experimental results show that through the image fusion required from wavelet transform image fusion method has more discrimination and much more quality improved. It is useful in a broad range of research filed. Key