Edge-preserved neural network model for image restoration
ABSTRACT This paper presents a combined approach for image restoration with edge-preserving regularization, subband coding, and artificial neural network. The edge information is detected from the source image as a prior knowledge to recover the details and reduce the ringing artifact of the subband coded image. The multilayer perceptron model is employed to implement the restoration of images. The main merit of the presented approach is that the neural network model is massively parallel with stronger robustness for transmission noise and parameter or structure perturbation, and it can be realized by very large scale integrated technologies for real-time applications. To evaluate the performance of the proposed approach, a comparative study with the set partitioning in hierarchical tree (SPIHT) has been made by using a set of gray-scale digital images. The experiment has shown that the proposed approach could result in considerably better performances compared with SPIHT on both objective and subjective quality for lower compression ratio subband coded image.
[Show abstract] [Hide abstract]
ABSTRACT: Grey target theory is a newly developed method in grey system theory. Calculating with grey target theory, pattern recognition, mode gradation and optimal mode selection can be performed. Oil samples from one auto press line and one 16V280ZJA diesel engine were analyzed by analytical ferrograph. For every wear particle in every analytical result, the size and quantity were quantified together. And some corresponding series have been established. Grey target theory was applied and a “bull's-eye” was constructed. By calculating the approaching degrees, wear modes have been graded, in conjunction with the running conditions of the press line and the diesel engine for given operation modes. The order of existing wear modes from severe to benign has also been assessed. The results agree with the experts’ analysis. This is the first time that grey target theory has been applied in oil monitoring for wear mode recognition.Wear 02/2006; 260(4-5-260):438-449. DOI:10.1016/j.wear.2005.02.085 · 1.86 Impact Factor
[Show abstract] [Hide abstract]
ABSTRACT: Image thresholding is the fundamental procedure in image processing. Meanwhile, edge information is a very useful and important image representation. A hierarchical multilevel thresholding method for edge information extraction using fuzzy entropy is presented in this paper. To realize multilevel thresholding fast and effectively, a tree structure is used to express the histogram hierarchy of an image. In each level of the tree structure, the image is segmented by three-level thresholding algorithm based on the maximum fuzzy entropy principle. In theory, the histogram hierarchy can be combined arbitrarily with multilevel thresholding. In order to evaluate the edge information extraction performance of multilevel thresholding methods, an edge similarity function is developed for according to the edge matching metric. Several images are employed to calculate their edge similarity coefficients. Experiments show that the proposed edge similarity coefficient is a valid one to measure the similarity between two image edge maps and it avoids the process effectively to obtain truth edge maps of images which can be realized only by labor statistics. To evaluate the performance of the proposed multilevel thresholding algorithm, the thresholded values of test images are calculated and compared using the proposed method, the Otsu and Kapur method, as well as edge similarity coefficients with the original images. The experimental results show that the proposed method spends less time to reach the better thresholds in edge similarity than existing multilevel thresholding methods.12/2011; 3(4). DOI:10.1007/s13042-011-0063-7