Research Of Infant Brain MR Image Enhancement Algorithm

In the process of imaging, due to various factors makes the image having different degree of degradation. Especially in the medical image processing, due to the characteristics of the human brain and the complex struction of the image, making the image having drawbacks existed.for example, Partial volume effection (PV effection), due to the movement of the patient’s posture leading to the boundary between the image of each organization rather vague; The noise of image, mainly because of the impact with the instruments and equipments; And gray uniform, due to the non-uniformity of the radiation field caused. Also, compared with the adult brain,the infant image having the lower image contrastion, the effection is more obvious for making the image more blurred. Therefore, this study mainly revolves around the infant brain image enhancement algorithms.The main contents are as follows:(1) On the basis of the traditional fractional differential, studying of the non-integer step fractional differential algorithm, The difference between the improved algorithm is using Lagrange non-integer step into fractional differential, the fractional differential with a single degree of freedom (differential order v), with improvements to the order (v) and unit step aliquot of copies (n) two degrees of freedom. Thus in the process of image enhancement, it will better to retain the continuity of the edge information, not appearing the edge’s fracture. Finally, using of the wavelet decomposition and reconstruction algorithm information, the image can be decomposed into low and high frequency information. Respectively, processing the high-frequency part and low-frequency part of the image, Compared with the traditional wavelet algorithm,the new algorithm has obvious advantages, such as the edge of the image can be enhanced obviously, whitish chunks will not happen and so on. Experiments show that the algorithm can enhanced infant brain image.(2) Based on the uneven of the infant image, combined with the image gradient information, the strengthen factors being proposed to add to the fractional differential algorithm. The new algorithm, combine with the gradient image and the fractional differential adaptive image enhancement algorithm. Using the actual infant brain images.The resulting image appropriate to adjust the brightness of the image non-uniformity problems; based on adaptive fractional differential algorithm when dealing with different images, does not requiring constant adjustment of the order of fractional differential problem, can save a lot of manual adjustments time, better to complete real-time image processing problems. The image with the actual operation, there may be a problem for the over-enhanced image, based on the above issues, this paper presents a blend of neighborhood Markov model algorithm for noise reducing, Markov model to calculate pixel neighborhood group energy,as a iterrative operating condition for the reduction of the noise. Each pixel in the image is detected and reduced the noise iterativly, to guarantee a good quality of the enhanced image.(3) Due to the frequency domain for image enhancement need not consider the spatial information of the image, reducing the time of computing. Enhancement algorithm based on frequency domain methods contain unsharp masking enhancements, discrete cosine transform (DCT) and wavelet transform. This article is dealing the uneven brightness of the image. Dividing it into blocks for processing, low-frequency coefficients of discrete cosine transform will be processed, the mainly application of the equations of motion is iterative model of bistable stochastic resonance. This image enhancement algorithm greatly increase the efficiency, thereby saving the entire algorithm computing time.