K-FCMNN based Brain MR Image Segmentation

Abstract:
Accurate brain tissue segmentation from magnetic resonance (MR) images is an essential step in quantitative brain image analysis. Segmentation of major brain tissues, including gray matter (GM), white matter (WM), and cerebrospinal fluid, from magnetic resonance (MR) images plays an important role in both clinical practice and neuroscience research. However, due to the non-uniform magnetic field or susceptibility effects, brain MR images may contain a smoothly varying bias field, which is also referred to as the intensity inhomogeneity or intensity non-uniformity. Although such modification improves the segmentation accuracy, it also dramatically increases the computational Complexity. Similarly, various kernel techniques have been used to improve the performance of clustering approaches. However, the clustering performed in a kernel space is generally very time consuming. To overcome this problem, the fuzzy local GMM (FLGMM) algorithm for brain MR image segmentation is used. The objective function of this algorithm is defined as the integration of the weighted GMM energy functions over the entire image. Thus local image data within the neighbourhood of each voxel follow GMM and fuzzy based decision system is used for the segmentation of the image. Thus we propose an algorithm K-FCMNN(K-Fuzzy Clustering Method for Nearest Neighbours).In this algorithm, the group of same datas are clustered in the given image and fuzzy approach is used for the segmentation of the image.Our results is applicable in both synthetic and clinical images show that the proposed algorithm can largely overcome the difficulties raised by noise, low contrast, and bias fields, and is capable of producing more accurate segmentation results than several state-of-the-art algorithms.

Effective Image Restoration Using Adaptive Contrast Enhancement and Bilateral Filter
Abstract:
Tone mapping (TM) is a method that maps high dynamic range (HDR) image to low dynamic range (LDR) image for display devices with limited dynamic range (DR). TM algorithms have been developed for reproducing the tone-mapped colour image, in which colour, contrast, and detail components are enhanced using luminance compression and colour reproduction by considering the human visual system or the local statistical characteristic. HDR image is captured using an HDR camera, which has high and low sensitivity sensors per pixel to increase DR. Noise is contained in HDR images, which are captured with high international organization for standardization (ISO) setting under the low light condition such as dim interior and night scene. Most conventional TM algorithms do not consider noise. HDR image contains both coarse-grain (low-frequency) and finegrain (high frequency) noise. To overcome the above Problem, we propose a noise reduction method and an adaptive contrast enhancement for local tone mapping (TM). The proposed local TM algorithm compresses the luminance of high dynamic range (HDR) image and decomposes the compressed luminance of HDR image into multi-scale subbands using the discrete wavelet transform. The proposed local TM algorithm decomposes an initial compressed luminance into multi-scale subbands using the discrete wavelet transform. The decomposed subbands are filtered using a bilateral filter (LL subband) and soft-thresholding( LH, HL, and HH subbands). And then, the local contrast is enhanced by an adaptive weight, which is derived from the luminance compression function with the color constraint. Finally, the color of the tone-mapped image is reproduced using an adaptive saturation control parameter. The Proposed approach out performs well for most set of the images.

Abstract: Image super-resolution (SR) reconstruction is the process of generating an image at a higher spatial resolution by using one or more low-resolution (LR) inputs from a scene.many methods for SR recon-struction have been proposed, which can broadly be grouped into three categories: interpolation-based methods, multiimage-based methods, and example-learning-basedmethods. the LR image patches in the test data set and the training data set is estimated to synthesize the desired HR image patches.However these algorithm depends heavily on a large training data set and is weak in generality.Similarly NE-based method is twofold: 1) The LRHR feature mappings cannot be established well because features in a high-dimensional space cannot be represented perfectly in a low-dimensional space, leading to some ambiguity between LRHR patch pairs; and 2) the strictly fixed neighbourhood size usually ends up with blurring effects, due to over- or under- fitting. To Overcome the above mention problem,we propose a sparse neighbor selection scheme for SR reconstruction. We first predetermine a larger number of neighbors as potential candidates and develop an extended Robust-SL0 algorithm to simultaneously find the neighbors and to solve the reconstruction weights. Recognizing that the -nearest neighbor( -NN) for re-construction should have similar local geometric structures based on clustering, we employ a local statistical feature, namely histograms of oriented gradients (HoG) of low-resolution (LR) image patches, to perform such clustering. By conveying local structural information of HoG in the synthesis stage, the -NN of each LR input patch is adaptively chosen from their associated subset, which significantly improves the speed of synthesizing the HR image while preserving the quality of reconstruction.