Driver fatigue problem is one of the important factors that cause traffic accidents. Therefore the vision-based driver fatigue detection is the most prospective commercial applications of HCI. However, it is a challen...
详细信息
ISBN:
(纸本)0769525210
Driver fatigue problem is one of the important factors that cause traffic accidents. Therefore the vision-based driver fatigue detection is the most prospective commercial applications of HCI. However, it is a challenging issue due to a variety of factors such as head and eyes moving fast, external illuminations interference and realistic lighting conditions, etc. This tends to significantly limit its scope of application. In this paper, we present an intelligent vehicle control based on driver fatigue detection. Firstly, the face is located using Haar algorithm and eye location is found with projection technique. After finding eye templates, we propose a new real time eye tracking method based on unscented Kalman filter. Thirdly, driver fatigue can be detected whether the eyes are closed over 5 consecutive frames using vertical projection matching. Finally, if driver fatigue is confirmed, the vehicle cruise control is start-up with slow speed, and maintains set slow speed such as 5 km/h. The experimental results show that intelligent vehicle control based on driver fatigue detection will be availability in traffic
Conventionally, manual operations that specify positions of feature points such as eyes and nose are needed when morphing is carried out for a face image. In this work, the feature points are therefore extracted by us...
详细信息
Conventionally, manual operations that specify positions of feature points such as eyes and nose are needed when morphing is carried out for a face image. In this work, the feature points are therefore extracted by using face area detection and a feature points decision methods to automate positional specification of feature points. As a result, the morphing of a face image can be carried out without manually specifying feature points. Face area detection is achieved by a threshold method using the YIQ color system. Feature points decision method extracts feature points by using a 3 layer perceptron type neural network (back-propagation). The attribute of the feature of eyes is defined to be a value of A in the color system LAB. In the same way, the attribute of feature points of the lip is defined as a value of B in the color system LAB. The extraction experiment of feature points was conducted from 120 face images by using the neural network, and the effectiveness of the present method was verified
Circular shortest path algorithms for polar objects segmentation have been proposed in the works of B. Appleton and C. Sun (2003, 2005) and C. Sun and S. Pallottino (2002) to address discrete case and extended in the ...
详细信息
Circular shortest path algorithms for polar objects segmentation have been proposed in the works of B. Appleton and C. Sun (2003, 2005) and C. Sun and S. Pallottino (2002) to address discrete case and extended in the work of B. Appleton and H. Talbot (2005) to the continuous domain for closed global optimal geodesic calculation. The best method up to date relies on a branch and bound approach and runs in O(u 1.6 v) on average while O(u 1.6 v) in worst case for a u times v discrete trellis warped in the direction of v. We propose an new algorithm called dichotomic multiple search (DMS) which finds the global minimum with a O(ulog 2 (u)v) worst case scenario complexity. Our algorithm relies on the fact that two minimal paths never cross more than once. This allows to sequentially partition the trellis in a dichotomic manner. Each computed circular minimal path with chosen starting point allows cutting the trellis into two sub trellis. The algorithm is then recursively applied on each sub trellis. Application to object segmentation is presented
Image retrieval methods aim to retrieve relevant images from an image database that are similar to the query image. The ability to effectively retrieve non-alphanumeric data is a complex issue. The problem becomes eve...
详细信息
Image retrieval methods aim to retrieve relevant images from an image database that are similar to the query image. The ability to effectively retrieve non-alphanumeric data is a complex issue. The problem becomes even more difficult due to the high dimension of the variable space associated with the images. Image classification is a very active and promising research domain in the area of image management and retrieval. In this paper, we propose a new image classification and retrieval scheme that automatically selects the discriminating features. Our method consists of two phases: (i) classification of images on the basis of maximum cross correlation and (ii) retrieval of images from the database against a given query image. The proposed retrieval algorithm recursively searches similar images on the basis of their correlation against a given query image from a set of registered images in the database. The algorithm is very efficient, provided that the mean images of all of the classes are computed and available in advance. The proposed method classifies the images on the basis of maximum correlation so that the images with more similarities and, hence, exhibiting maximum correlation with each other are grouped in the same class and, are retrieved accordingly.
While increasing amounts of complex information are becoming available on the web, there is, beyond keywordbased search and listing of results, a paucity of user interface paradigms and implementations that support in...
详细信息
While increasing amounts of complex information are becoming available on the web, there is, beyond keywordbased search and listing of results, a paucity of user interface paradigms and implementations that support interaction, exploration, and assimilation of information. This paper describes our design of a novel framework to address this deficiency. The proposed framework supports both direct search behavior as well as more exploratory search strategies through multiple-perspective visualization and interaction with search results. The approach is developed around the twin themes of supporting data context and facilitating effective interactions between users and data. The system supports data context through determination of semantic correlations between web pages and extraction of the spatio-temporal data contained therein. A multipleperspective environment is then used to display semantic and spatio-temporal relationships as well as to provide intuitive views of the data, specifically through web page thumbnail, map, and timeline modules. The environment supports direct interactions with the data through a reflective interface by which user selections in any one panel highlight the corresponding information in other panels. In this environment, visual cues and explicit facilities to model space and time aid in recognition, querying, and exploration of information as well as in representation and reasoning with complex relationships (such as spatio-temporal, causal, evolutionary) in the data. Experimental studies of a quantitative and qualitative nature demonstrate the efficacy of the system in facilitating both information extraction and discovery.
In this paper, we present a novel graph theoretic approach to the problem of document-word co-clustering. In our approach, documents and words are modeled as the two vertices of a bipartite graph. We then propose isop...
详细信息
In this paper, we present a novel graph theoretic approach to the problem of document-word co-clustering. In our approach, documents and words are modeled as the two vertices of a bipartite graph. We then propose isoperimetric co-clustering algorithm (ICA) - a new method for partitioning the document-word bipartite graph. ICA requires a simple solution to a sparse system of linear equations instead of the eigenvalue or SVD problem in the popular spectral co-clustering approach. Our extensive experiments performed on publicly available datasets demonstrate the advantages of ICA over spectral approach in terms of the quality, efficiency and stability in partitioning the document-word bipartite graph.
Based on a new distance, a novel noise-resistant fuzzy clustering algorithm, called alternative noise clustering (ANC) algorithm, is proposed. ANC is an extension of the noise clustering (NC) algorithm proposed by Dav...
详细信息
Based on a new distance, a novel noise-resistant fuzzy clustering algorithm, called alternative noise clustering (ANC) algorithm, is proposed. ANC is an extension of the noise clustering (NC) algorithm proposed by Dave. By replacing the Euclidean distance used in the objective function of NC algorithm, a new distance (non-Euclidean distance) is introduced in NC algorithm. Based on robust statistical point of view and influence function, the non-Euclidean distance is more robust than the Euclidean distance. So the ANC algorithm is more robust than the NC algorithm. Moreover, with the new distance ANC can deal with noises or outliers better than NC and fuzzy c-means (FCM). The better performance of the proposed algorithm is shown by performing experiments on data sets
The two-dimensional linear discriminant analysis (2DLDA) is one of the most successful face recognition methods. However, it cannot be directly applied to the face recognition where only one sample image per person is...
详细信息
The two-dimensional linear discriminant analysis (2DLDA) is one of the most successful face recognition methods. However, it cannot be directly applied to the face recognition where only one sample image per person is available for training. In this paper, we present a new method based on 2DLDA to deal with the single training sample problem. The method derive a set of sub-images from a single face image by sampling, therefore obtaining multiple training samples for each class, and then apply 2DLDA to the set of newly produced samples. The proposed algorithms are compared with both the E(PC) 2 A algorithm and the SVD perturbation algorithm which is proposed for addressing the single training sample problem. Experimental results on the ORL face database show that the proposed approach is feasible and has higher recognition performance than E(PC) 2 A and SVD perturbation algorithms
This paper proposes a novel method for estimating the geospatial trajectory of a moving camera. The proposed method uses a set of reference images with known GPS (global positioning system) locations to recover the tr...
详细信息
This paper proposes a novel method for estimating the geospatial trajectory of a moving camera. The proposed method uses a set of reference images with known GPS (global positioning system) locations to recover the trajectory of a moving camera using geometric constraints. The proposed method has three main steps. First, scale invariant features transform (SIFT) are detected and matched between the reference images and the video frames to calculate a weighted adjacency matrix (WAM) based on the number of SIFT matches. Second, using the estimated WAM, the maximum matching reference image is selected for the current video frame, which is then used to estimate the relative position (rotation and translation) of the video frame using the fundamental matrix constraint. The relative position is recovered up to a scale factor and a triangulation among the video frame and two reference images is performed to resolve the scale ambiguity. Third, an outlier rejection and trajectory smoothing (using b-spline) post processing step is employed. This is because the estimated camera locations may be noisy due to bad point correspondence or degenerate estimates of fundamental matrices. Results of recovering camera trajectory are reported for real sequences
This paper describes an efficient feature selection method that quickly selects a small subset out of a given huge feature set; for building robust object detection systems. In this filter-based method, features are s...
详细信息
This paper describes an efficient feature selection method that quickly selects a small subset out of a given huge feature set; for building robust object detection systems. In this filter-based method, features are selected not only to maximize their relevance with the target class but also to minimize their mutual dependency. As a result, the selected feature set contains only highly informative and non-redundant features, which significantly improve classification performance when combined. The relevance and mutual dependency of features are measured by using conditional mutual information (CMI) in which features and classes are treated as discrete random variables. Experiments on different huge feature sets have shown that the proposed CMI-based feature selection can both reduce the training time significantly and achieve high accuracy.
暂无评论