Search

To resolve the difficult problem of soft tissue segmentation of MRI images of the segmentation of the multi target region of interest in the MRI images with complex targets and fuzzy boundary, this paper proposed a novel MRI image segmentation method, based on anisotropic diffusion and spatial fuzzy C-means clustering (SFCM). It preprocessed the images using the nonlinear and anisotropic diffusion, resolving the problem of weakening the image details while removing the noise. It designed space function combining with the neighborhood space, improving traditional FCM objective function. It used the spatial information of the image to achieve the accurate classification of every object in image was an effective solution to the isolated area correctly classified. After that, it obtained complete and continuous segmented regions. Finally, it utilized the fitting curve of histogram to initialize the classification number and the initial cluster centers, accelerating the algorithm iterative to the optimal solution, and also reducing the run time. The experiments show that the possibility of find a best solution is improved by introducing the method of MRI image segmentation method based on anisotropic diffusion and spatial FCM, so as to the processing of MRI images segmentation which has overlapped grayscale, discontinuous objects and fuzzy boundary. [ABSTRACT FROM AUTHOR]

To improve predicting accuracy further, this paper modified original Pareto dominance relation among the candidate solutions, and proposed a new method named d-Pareto dominance prediction using nearest neighbor. Combined with the characteristics of multi-objective optimization, it described the framework of d-Pareto and the conclusion that a d-Pareto dominance prediction could obtain a lower average prediction error rate comparing with Pareto demonstrated dominance prediction. Besides, it also explored the interaction between d-Pareto dominance prediction and multi-objective evolutionary algorithms. Experiments on several classic MOPs were conducted and the simulation results show that prediction of d-Pareto dominance is feasible and effective. [ABSTRACT FROM AUTHOR]

For quantitative-based and relationship-based embedding methods can not resist brightness enhancement and common attacks respectively, this paper proposed a hybrid quantization and relationship color image watermarking algorithm. Firstly, it subdivided each component of the color image to non-overlapping blocks of size 8x8 and performed 1-level discrete wavelet transform (DWT) for each selected block by using a key, and then did singular value decomposition (SVD) on the low and high frequency sub-band, respectively. Finally, the biggest singular value of low and high frequency sub-band exploited quantification and relationship method to embed preprocessed watermarking, respectively. Experimental results show that the proposed algorithm has low complexity and good transparency. Compared with other algorithms, it has better robustness performance. [ABSTRACT FROM AUTHOR]