Catálogo de publicaciones - libros

Geometric constraint problem is equivalent to the problem of solving a set of nonlinear equations substantially. Nonlinear equations can be solved by classical Newton-Raphson algorithm. Path tracking is the iterative application of Newton-Raphson algorithm. The Homotopy iteration method based on the path tracking is appropriate for solving all polynomial equations. Due to at every step of path tracking we get rid off the estimating tache, and the number of divisor part is less, so the calculation efficiency is higher than the common continuum method and the calculation complexity is also less than the common continuum method.

Combining the reliable edge feature points and area similarity, the fast stereo matching algorithm using edge traction was presented. First, find valid disparity set of feature points and traverse combinations of adjacent points’ disparities, obtain the valid disparity set of featureless points using dynamic program, then, generate the initial sparse disparity space using area similarity. The algorithm reduces the computation complexity of disparity space and decreases the possibility of mismatching illusion. Under the uniqueness constraint, integral dense disparity map and occlusion area can be obtained by collision detection. Experiment on real visual images is performed to verify the feasibility and effectiveness of this algorithm.

In this paper, we present a fully automatic digital chroma-keying system, which is based on the integration of color image segmentation algorithm and improved alpha estimation technique. Chroma-keying is a critical technology in virtual studio system. When used with cluttered background, it calls for much intelligence. According to the characteristics of frame images in the target application, a hybrid color image segmentation algorithm is put forward, which makes good use of both chromatic and luminance information. Then, refinement measures are further taken to deal with the color distribution in the neighborhood of the boundary through modified Ruzon-Tomasi alpha estimation algorithm. In contrast to the previously reported methods, our system needs no human interaction in the whole procedure. Experimental results on China sports lottery TV programs show that the proposed fully automatic keying system is viable and can be applied to the real program post production process of TV stations.

A new algorithm of recognition of continuous deformation of the images with multi-gray-levels is put forward in this paper, which the following steps are made: the fist is the adoption of griddings procedures, the other is the classification of continuous deformation into continuous deformation with preserving topological structure and continuous deformation with non-preserving topological structure, and introduction of a new method approximate identifying continuous deformation. The main characteristics of this algorithm are the flexible modulations between accuracy to calculate and time to process, and so the needs from different applications be satisfied easily. Finally, a few examples are given to test the versatility of the techniques, from which it is verified that the algorithm developed here exhibits good performance.

The efficiency of pattern recognition depends heavily on that if feature extraction and selecting are effective. Complicated image such as medical image and remote sensing image, belong to image with natural textures, this kind of image is always of high resolution, with many layers of gray degree, and a very intricate shape structure. Because there are no obvious shapes, but only distributions of some gray degrees and colors in these images, so for them, there are no good methods yet for feature extraction and region recognition. In this paper, based on information augmentation and kinetics, we present a learning algorithm, which can be used to do region classification of the above-mentioned images with natural textures. We applied our algorithm to recognition of image with natural textures and obtained a good result.

Two new methods for combination feature extraction are proposed in this paper. The methods are based on the framework of CCA in image recognition by improving the correlation criterion functions. Comparing with CCA methods, which can solve the classification of high-dimensional small size samples directly, being independent of the total scatter matrix singularity of the training simples, and the algorithms’ complexity can be lowered. We prove that the essence of two improved criterion functions is partial least squares analysis (PLS) and multivariate linear regression (MLR). Experimental results based on ORL standard face database show that the algorithms are efficient and robust.

Median filter is well known for removing impulsive noise and preserving edges. Repeatedly filtering of any one-dimensional signal with a median filter will produce a root signal. Any impulses in the input signal will be removed by sufficient number of passes of median filter, where any root like features in the input signal will be preserved. A signal of finite length will be filtered to a root signal after a finite number of passes of a median filter of a fixed window, results in the convergence of the signal. In this paper, root signal and its properties are analyzed for One-dimensional signal. Adaptive length median filter, weighted median filter, FIR hybrid median filter and Linear combination of weighted median filter have been taken and their root signals are obtained. Their performances are analyzed by determining Power spectrum density, Mean square error and Signal to noise ratio.

A fuzzy-rough set model is presented based on the extension of the classical rough set theory. The continuous attributes are fuzzified. The indiscernibility relation in classical rough set is extended to the fuzzy similarity relation. Then an inductive learning algorithm based on fuzzy-rough set model (FRILA) is proposed. Finally, with comparison to the decision tree algorithms, the effectiveness of the proposed method is verified by an example.

Due to the shortage of rough implication in [4] ∼ [6], rough set and rough implication operators are redefined by using interval structure in [7], the shortages have been e improved. We have investigated the characteristics of the rough implication, and also point out that the good logic property of the rough implication in [7]. In this paper, we will study the algebraic properties of the rough implication in depth.

Based on the definition of , this paper discusses detailedly the conditions on which the satisfy the information boundedness principle in fuzzy reasoning, and gets the corresponding conclusion: when fuzzy propositions have positive measuring errors for their membership grades, satisfy the information boundedness principle only if they are rejecting or restraining correlative; when they have negative ones, the operators satisfy the principle only if they are restraining correlative. This conclusion has important directive meaning for how to give the value of the in practical control application.