Catálogo de publicaciones - libros

As an effective method of background subtraction, codebook model suffers from unacceptable false negative detection rate in many situations due to its quantization criterion. In this paper, we propose an improved codebook model to solve this problem. Instead of using the original quantization criterion, we quantize the temporal series of the observations at a given pixel into codewords based on the Gaussian distribution assumption. We have performed this approach in our surveillance system for outdoor scenes and achieved excellent detection results.

A new Fourier-wavelet volume rendering algorithm (FWVR) in frequency domain is presented and corresponding preprocessing is given in this paper. After the wavelet transforming and obtaining slice plane, FWVR finish the volume rendering image with a wavelet-based X-ray rending method in the frequency domain. FWVR comprises and unifies the advantages of recently introduced Fourier domain volume rendering techniques and wavelet based volume rendering. For the algorithm is based on the CT imaging and wavelet transformation, the time complexity of this method is O( N ^2log N ) which is reduced apparently. Experiment of medicine volume data shows that this method is efficient. Finally, the potential application of FWVR is introduced in network rendering for saving transmission time.

In this paper, we present a new method of SAR imagery compression based on multiscale autoregressive moving average (MARMA) models. We use the multiscale representation as the cornerstone of the modeling process, and construct the MARMA models of SAR image. We then predict the initialized image data using these multiscale models. Next we compress image data through coding the residual image. Extension simulations have proven that the proposed method achieves high compression radios with impressive image quality.

When converting image sequence to 3D, several entries of the matrix have not been observed by occlusions and other entries have been perturbed by the influence of noise. In this paper, we propose a method for fitting geometrically missing data in noisy observation matrix with iterative SVD factorization. The main idea of the proposed algorithm is that the orientation and distance of noisy vector ca be handled directly by geometrical properties between 2D image plane and 3D error space under noise distribution. To confirm the recoverability of missing data, we carry out the experiments for synthetic and real sequences. The results in practical situations demonstrated with synthetic and real video sequences verify the efficiency and flexibility of the proposed method.

The physical flame characteristics in the burning process are analyzed in detail. Combined with static and dynamic flame features, a flame detection algorithm based on multi-feature fusion is designed to quickly recognize fire flame using the theory of the degree of belief. The experiments show that the processing rate of the algorithm is as high as 25 frames/s without any manual adjustments. In complex situations, the algorithm remains with stronger anti-jamming ability and higher accuracy, for example, the suspected probability of street lamps and automotive lamps is lower than 10%. From comparing results, the video-based detection algorithm has the good capabilities.

A series of high order Mandelbrot and Julia fractal images of complex map $f:z\leftarrow{\mathcal{}z}^n+c(n>2,n\in{N}) $ are constructed by using escape time algorithm. Topology invariance and the relation between period bud distribution and mapping orders are found through lots of computer mathematic experiments. The escape time algorithm is improved on basis of the rotation symmetric property of Manderlbrot and Julia sets, and then a rotation escape time algorithm is established. The high-order Mandelbrot and Julia sets fractal images are generated by using Web oriented Java Applet. Therefore the complexity of computer simulation under complex conditions is greatly reduced. The application of Java technology provides an Internet-based distribution platform for chaos-fractal theory study.

Most texture-based image retrieval system just consider an original image of coarseness, contrast and roughness, actually there are many texture information in the edge image. In this paper, a method combining both edge information and gray level co-occurrence matrix properties is proposed to improve the retrieval performance. The proposed method gives encouraging results when comparing its retrieval performance to that of the Yao’s method, in the same image database.

Many researches on path planning and obstacle avoidance for the fundamentals of mobile robot have been done. Although many solutions help finding a optimal path, those can be applied to real world only under constrained condition, which means that it is difficult to find a universal algorithm. Moreover, a complicated computation to obtain an optimal path induces the time delay so that a robot can not avoid moving obstacles. In this paper, we propose the algorithm of path planning and obstacle avoidance using Random Access Sequence(RAS) methodology. In the proposed scheme, the cell decomposition is make first and cell neighbors are assigned as sequence code, then the path with minimum length is selected.

Based on multiwavelet transform and the clonal selection theory in the natural immune system, a novel pixel-level multifocus image fusion optimization algorithm is presented in this paper. Source images are first decomposed into low-frequency coarse information and high-frequency detail information via discrete multiwavelet transform. The high-frequency detail information adopts the absolute-values maximum selection. And then the immune clonal selection is introduced to optimize the weights of fusing the low-frequency four coarse subbands adaptively and separately. Image fusion performances of Daubechies-4 (Db4) scalar wavelet, Geronimo, Hardin and Massopust (GHM) multiwavelets and Chui and Lian (CL) multiwavelets are compared quantitatively, which have the same approximation order. Experimental results show that the proposed image fusion algorithm have clear edges, abundance details and few artificial artifacts.

Two-dimensional computations of the propagation of the explosion wave in a mixture of hydrogen and oxygen are performed using WENO finite difference scheme with fifth order accuracy and two-step chemical reaction model. Numerical results show that obstacles setting repeatedly along the path of the flame propagation process obviously can accelerate the flame. The mechanism of the accelerating effect is attributed to the positive feedback of the turbulence region induced by obstacles for combustion process.