Catálogo de publicaciones - libros

An application for content-based annotation and retrieval of videos can be found in the sport domain, where videos are annotated in order to produce short summaries for news and sports programmes, edited reusing the video clips that show important highlights and the players involved in them. The problem of detecting and recognizing faces in broadcast videos is a widely studied topic. However, in the case of sports videos in general, and soccer videos in particular, the current techniques are not suitable for the task of face detection and recognition, due to the high variations in pose, illumination, scale and occlusion that may happen in an uncontrolled environment.

In this paper we present a method for face detection and recognition, with associated metric, that copes with these problems. The face detection algorithm adds a filtering stage to the Viola and Jones Adaboost detector, while the recognition algorithm exploits local features to describe a face, without requiring a precise localization of the distinguishing parts of a face, and the set of poses to describe a person and perform a more robust recognition.

The detection of commercials in news video has been a challenging problem because of the diversity of the production styles of commercial programs. In this paper, the authors present a novel algorithm for the detection of commercials in news program. By the method suggested, firstly shot transition detection and anchorman shot recognition are conducted, then clustering analysis is employed to label commercial blocks roughly, finally the accurate boundaries of the commercials are located by analyzing the average duration of preceding and subsequent shots and the visual features of the shots, such as color, saturation and edge distribution. The experiment results show that the proposed algorithm is effective with high precision.

The use of salient regions is an increasingly popular approach to image retrieval. For situations where object retrieval is required and where the foreground and background can be assumed to have different characteristics, it becomes useful to exclude salient regions which are characteristic of the background if they can be identified before matching is undertaken. This paper proposes a technique to enhance the performance of object retrieval by filtering out salient regions believed to be associated with the background area of the images. Salient regions from background only images are extracted and clustered using descriptors representing the salient regions. The clusters are then used in the retrieval process to identify salient regions likely to be part of the background in images containing object and background. Salient regions close to background clusters are pruned before matching and only the remaining salient regions are used in the retrieval. Experiments on object retrieval show that the use of salient region background filtering gives an improvement in performance when compared with the unfiltered method.

A novel method for moving video objects recognition is presented in this paper. In our method, support vector machine (SVM) is adopted to train the recognition model. With the trained model, the moving video objects can be recognized based on the shape features extraction. Comparing with the traditional methods, our method is faster, more accurate and more reliable. The experimental results show the competitiveness of our method.

In this paper, we propose an EM based Multiple-Instance learning algorithm for the image classification and indexing. To learn a desired image class, a set of exemplar images are selected by a user. Each example is labeled as conceptual related (positive) or conceptual unrelated (negative) image. A positive image consists of at least one user interested object, and a negative example should not contain any user interested object. By using the proposed learning algorithm, an image classification system can learn the user’s preferred image class from the positive and negative examples. We have built a prototype system to retrieve user desired images. The experimental results show that for only a few times of relearning, a user can use the prototype system to retrieve favor images from the WWW over Internet.

In this paper, we propose a scheme of personal authentication using palm vein. The infrared palm images which contain the palm vein information are used for our system. Because the vein information represents the liveness of a human, this system can provide personal authentication and liveness detection concurrently. The proposed system include: 1) Infrared palm images capture; 2) Detection of Region of Interest; 3) Palm vein extraction by multiscale filtering; 4) Matching. The experimental results demonstrate that the recognition rate using palm vein is good.

A novel Support Vector Machine (SVM) face recognition method using optimized Gabor features is presented in this paper. 200 Gabor features are first selected by a boosting algorithm, which are then combined with SVM to build a two-class based face recognition system. While computation and memory cost of the Gabor feature extraction process has been significantly reduced, our method has achieved the same accuracy as a Gabor feature and Linear Discriminant Analysis (LDA) based multi-class system.

This paper presents an efficient matching algorithm for the palmprint identification system. Line segments are extracted from an image as primitives. Each local structure is represented by a set of pair-wise angle relationships, which are simple, invariant to translation and rotation, robust to end-point erosion, segment error, and sufficient for discrimination. The Earth Mover’s Distance (EMD) was proposed to match the pairwise relative angle histograms. EMD not only supports partial matching but it also establishes a neighbouring relationship in the lines information during the matching process. The system employs low-resolution palmprint images captured by normal digital camera and achieves higher identification accuracy with lower time complexity.

Lip movement of a speaker conveys important visual speech information and can be exploited for Automatic Speech Recognition. While previous research demonstrated that visual modality is a viable tool for identifying speech, the visual information has yet to become utilized in mainstream ASR systems. One obstacle is the difficulty in building a robust visual front end that tracks lips accurately in a real-world condition. In this paper we present our current progress in addressing the issue. We examine the use of color information in detecting the lip region and report our results on the statistical analysis and modeling of lip hue images by examining hundreds of manually extracted lip images obtained from several databases. In addition to hue color, we also explore spatial and edge information derived from intensity and saturation images to improve the robustness of the lip detection. Successful application of this algorithm is demonstrated over imagery collected in visually challenging environments.

Principal components analysis (PCA) is a popular linear feature extractor to unsupervised dimensionality reduction, and found in many branches of science including-examples in computer vision, text processing and bioinformatics, etc. However, axes of the lower-dimensional space, i.e., principal components, are a set of new variables carrying no clear physical meanings. Thus, interpretation of results obtained in the lower-dimensional PCA space and data acquisition for test samples still involve all of the original measurements. To select original features for identifying critical variables of principle components, we develop a new method with k-nearest neighbor clustering procedure and three new similarity measures to link the physically meaningless principal components back to a subset of original measurements. Experiments are conducted on benchmark data sets and face data sets with different poses, expressions, backgrounds and occlusions for gender classification to show their superiorities.