Reserch On Infant Brain MR Image Enhancement Algorithm Based On Improved Fractional Order Differential

With the research and development of computer-aided diagnosis,medical imaging has become the most commonly used and important tool for doctors to diagnose and treat diseases.Nuclear magnetic resonance(MR)is one of the safe and effective imaging technology,and it is also the most common tool for the diagnosis of brain diseases in infants and young children.The problem of fuzzy,intensity inhomogeneity,artifacts and offset field image of infants with brain MR not only affects doctor’s diagnosis and treatment,but also causes difficulties in the subsequent automatic segmentation and brain development analysis.Therefore,this thesis proposes three image enhancement algorithms for infant brain MR images.The main work of this thesis is:(1)The image enhancement algorithm based on improved fractional order differential and non-local means is studied.We use the Otsu algorithm and the local average gradient to divide the image into strong edge,weak edge,strong texture,weak texture and smooth region,and determine the fractional order of each region adaptively.Then the image roughness is added to get the initial fractional order.The non neighborhood block made up of the initial order and the image gray level of non-neighborhood is filtered to obtain the final order.Experimental results show that the proposed algorithm has good performance in enhancing texture and suppressing noise.(2)The image enhancement algorithm based on multi-direction adaptive fractional order differential is studied.In this thesis,we first determine the number of directions at each point in the image,and use the Bresenham linear algorithm to determine the position and gray value of each direction of center point.The similarity function is then used to determine the length in each direction,and the fractional order in each direction is determined.Finally,fractional order differential filtering is applied to the pixels in each direction.Experiments show that the proposed algorithm can enhance the texture and preserve the structural information of the image.(3)The image enhancement algorithm based on morphological component analysis and improved fractional order differential is studied.In the framework of the morphological component analysis,the image is decomposed by the characteristics of roughness,contrast,direction and linear correlation,and each component is.manipulated separately.Finally,the weighted sum of the manipulated components is added to obtain the final enhanced image.The common image segmentation algorithms are used to segment the enhancement results.It is proved that the algorithm can improve the image segmentation accuracy and is the basis of computer aided diagnosis.