High Resolution Medical Image Reconstruction Based On Image Fusion

With the continuous development and progress of science and technology, the application of computer is more and more common, especially in the field of medical image processing. Image denoising technique is the part of image preprocessing before any other image processing operations, it has great influence on the results of image processing, especially in the field of medicine, the quality of processed image is very important to the clinical diagnosis result. One of the main factors to measure the quality of the image is the resolution of the image. In the process of image acquisition, the image quality will be reduced because of the changes of the parameters of the imaging equipment, the environment temperature and other external factors. A common phenomenon is the image has blocks of fuzzy. In this paper, according to this phenomenon, the high resolution medical image is reconstructed by wavelet transform and compressed sensing algorithm.Firstly, the image denoising algorithm is introduced. Several widely used classical image denoising algorithms are introduced respectively, such as bilateral filtering image denoising algorithm, the anisotropic diffusion denoising algorithm and mean filtering algorithm and non local mean image denoising algorithm. Then, this paper improves the shortcomings of the kernel function which is used to calculate the weighted coefficient in the non local mean denoising algorithm. A new weighted kernel function is proposed by combing the Bisquare weighted kernel function and cosine kernel function. And it is applied to the calculation of the weighted coefficient of the denosing process of the medical image.Secondly, this paper introduces some existing image fusion algorithms and the evaluation criteria of image fusion. Emphasis is put on the analysis of the theory of compressed sensing and the application of the theory in medical image fusion, and some fusion rules which are used in image fusion are analyzed and compared. And according to the characteristics of medical images, improving the rule of coefficient fusion based on average which is used in traditional algorithm. An improved compressive sensing fusion algorithm based on region variance is proposed for medical images.Then, this paper introduces the principle of the image fusion method based on wavelet transform as well as some common fusion methods, and analyzes the advantages and disadvantages of various fusion algorithms. Finally, according to the characteristics of the wavelet coefficients of the medical image decomposition, an image fusion method based on the combination of the spatial frequency and the regional variance is proposed. For the low frequency wavelet coefficients, the spatial frequency is used as the selection criterion of the fused coefficients. For the high-frequency coefficients, first distinguish the high frequency coefficients of different scales by direction. Then, in view of the high-frequency coefficients in different directions, adopting fusion rules based on the combination of regional variance and weighted proportional.Finally, using the image user interface(GUI) of MATLAB to design the high resolution medical image reconstruction software. The software mainly includes two parts:(1) image denoising;(2) image fusion.