Catálogo de publicaciones - libros

In this paper, we propose a new registration algorithm, which can provide more exact criteria for deciphering brain diseases. At the first stage, our algorithm divides the areas of the brain structures to extract their features. After calculating contour and area information, the grouping step is performed. At the next stage, the brain structures are precisely classified with respect to the shape of cerebrospinal fluid and the volume of brain structures. These features are finally integrated into a knowledge base to build up a new standard atlas for normal brain MR images. Using this standard atlas, we perform the registration process after extracting the brain structures from the MR image to be compared. Finally, we analyze the registration results of the normal and abnormal MR images, and showed that the exactness of our algorithm is relatively superior to the previous methods.

We present a parameterized representation of virtual organs for surgery simulation purpose. Random 3D input mesh are parameterized and resampled into a regular 2D parameterized model. With this parameterized representation, a high resolution 3D organ mesh can be reconstructed and deformed interactively with a simple and fast free-form deformation method. The amount of deformation and force feed-back can be calculated rapidly. Therefore, haptic rendering can be achieved. In addition, the parameterized mesh can be used to handle collision detection and the contact between multi-objects in an efficient way. With the parameterized mesh, realistic visual and haptic rendering can be provided for interactive surgery simulation.

A new method of content-based image retrieval is presented that uses the color co-occurrence matrix that is adaptive to the classification characteristics of the image blocks. In the proposed method, the color feature vectors are extracted according to the characteristics of the block classification after dividing the image into blocks with a fixed size. The divided blocks are then classified as either luminance or color blocks depending on the average saturation of the block in the HSI (hue, saturation, and intensity) domain. Thereafter, the color feature vectors are extracted by calculating the co-occurrence matrix of a block average intensity for the luminance blocks and the co-occurrence matrix of a block average hue and saturation for the color blocks. In addition, block directional pattern feature vectors are extracted by calculating histograms after directional gradient classification of the intensity. Experimental results show that the proposed method can outperform conventional methods as regards a precision and a feature vector dimension.

Despite the rapid growth of wallpaper image downloading service in the mobile contents market, users experience high levels of frustration in searching for desired images, due to the absence of intelligent searching aid. Although Content Based Image Retrieval is the most widely used technique for image retrieval in the PC-based system, its application in the mobile Web environment poses one major problem of not being able to satisfy its initial query requirement because of the limitations in user interfaces of the mobile application software. We propose a new approach, so called a CF-fronted CBIR , where Collaborative Filtering (CF) technique automatically generates a list of candidate images that can be used as an initial query in Content Based Image Retrieval (CBIR) by utilizing relevance information captured during Relevance Feedback. The results of the experiment using a PC-based prototype system verified that the proposed approach not only successfully satisfies the initial query requirement of CBIR in the mobile Web environment but also outperforms the current search process.

This paper describes how the covariance information in MPEG video data can be incorporated into a distance measure and applies the resulting divergence measure to video content classification problems. The divergence measure is adopted into two different clustering algorithms, the Centroid Neural Network (CNN) and the Gradient Based Fuzzy c-Means (GBFCM) for MPEG video data classification problems, movie or sports. Experiments on 16 MPEG video traces show that the divergence measure with covariance information can decrease the False Alarm Rate (FAR) in classification as much as 46.6% on average.

To improve the effectiveness and efficiency of CBIR systems, in this paper, we present a novel IR scheme with multiple features, the spatial color and edge percentage features, derived by way of moment-preserving edge detection. Put the above two features together, we come by an effective and efficient IR system. Experimental results show that the proposed method outperforms other similar methods in terms of accuracy and retrieval efficiency.

Multimedia Document (MMD) such as Web Page and Multimedia cyclopedias is composed of media objects of different modalities, and its integrated semantics is always expressed by the combination of all media objects in it. Since the contents in MMDs are enormous and the amount of them is increasing rapidly, effective management of MMDs is in great demand. Meanwhile, it is meaningful to provide users cross-media retrieval facilities so that users can query media objects by examples of different modalities, e.g. users may query an MMD (or an image) by submitting a audio clip and vice versa. However, there exist two challenges to achieve the above goals. First, how can we represent an MMD and fuse media objects together to achieve Cross-index and facilitate Cross-media retrieval? Second, how can we understand MMD semantics? Taking into account of the two problems, we give the definition of MMD and propose a manifold learning method to discover MMD semantics in this paper. We first construct an MMD semi-semantic graph (SSG) and then adopt Multidimensional scaling to create an MMD semantic space (MMDSS). We also propose two periods’ feedbacks. The first one is used to refine SSG and the second one is adopted to introduce new MMD that is not in the MMDSS into MMDSS. Since all of the MMDs and their component media objects of different modalities lie in MMDSS, cross-media retrieval can be easily performed. Experiment results are encouraging and indicate that the performance of the proposed approach is effective.

A system for multimedia retrieval and summarization in a ubiquitous environment is presented. Hierarchical clustering of data from pressure-based floor sensors is followed by video handover to retrieve video sequences showing the movement of each person in the environment. Audio handover is implemented to dub these sequences. Several methods for extracting key frames from the video sequences were implemented and evaluated by experiments. An adaptive spatio-temporal sampling algorithm based on the rate of footsteps yielded the best performance. The measured accuracy of key frame extraction within a difference of 3 seconds is approximately 80%. The system consists of a graphical user interface that can be used to retrieve video summaries interactively using simple queries.