Research On Bio-medical Image Denoising And Segmentation Algorithm

Bio-medical image processing has been paid much attention to in the field of computer vision and pattern recognition in recent decades. Bio-medical image which has the feature of intuition,image,information and so on, is easy to observe and has play an important role in medical diagnosis and experimental research. So bio-medical image acquisition, processing and analysis become cutting-edge issues. Based on bio-medical image segmentation, the image contents can be better analyzed, which will assist doctors to diagnose correctly and make up the medical programs. However, due to bio-medical image with low resolution,noise,large differences in the characteristics of image features, the segmentation of such images become much difficult. As a result, the main work of this paper as follows:(1) The traditional image denoising models, such as mean filter, median filter, winner filter and partial differential equations filter and so on, have the difficulty in denoising and keeping the details. The paper introduces the non-local means denoising model, which can consider the spatial relation ships and similarities between gray at the same time to overcome the shortcomings of coexistence of denoising and details. But this model has the disadvantage about time efficiency while searching the similar window. In order to solve this problem, the paper proposes an improved non-local means denoising method which chooses denoising method selectively. In the detail, it selects the non-local means denoising method to maintain good benefits; On the contrary, in the homogeneous areas, it selects Mean Shift denoising method to overcome the time disadvantage of low efficiency. Experimental results show that the method can improve the efficiency of the denoising of the time.(2) The traditional FCM model only considers gray level information while ignoring the spatial relationship between pixels, so it is sensitive to noise. In order to overcome the limitation of FCM model, the paper proposes an anisotropic fuzzy clustering segmentation model. The new FCM model can reduce the effect of the noise and contain the information of beam structure regions and corner regions. This paper apply it to virtual human brain image segmentation model and experiments on the segmentation of brain magnetic resonance images show this model can attain better effect in image segmentation.