International Journal of Pattern Recognition and Artificial Intelligence

Similarity-based retrieval from shape databases typically employs a pairwise shape matcher and one or more indexing techniques. In this paper, we focus specifically on the design of a pairwise matcher for retrieval of 2-D shape contours. In the past, the matchers used for the one-to-many problem of shape retrieval were often designed for the problem of matching an isolated pair of shapes. This approach fails to exploit two characteristics of the one-to-many matching problem that distinguish it from the one-to-one matching problem. First, the output of shape retrieval systems tends to be dominated by matches to relatively similar shapes. In this paper, we demonstrate that by not expending computational resources on unneeded accuracy of matching, both the speed and the accuracy of retrieval can be increased. Second, the shape database is a large statistical sample of the population of shapes. We introduce a probabilistic model for exploiting that statistical knowledge to further increase retrieval accuracy. The model has several benefits: (1) It does not require class labels on the database shapes, thus supporting unlabeled retrieval. (2) It does not require feature independence. (3) It is parameter-free. (4) It has a fast runtime implementation. The probabilistic model is general and thus potentially applicable to other one-to-many matching problems.

The sprite generator introduced in MPEG-4 blends frames by averaging, which will make places, that are always occupied by moving objects, look blurred. Thus, providing segmented masks for moving objects is suggested. Several researchers have employed automatic segmentation methods to produce moving object masks. Based on these masks, they used a reliability-based blending strategy to generate sprites. Since perfect segmentation is impossible, some ghost-like shadows will appear in the generated sprite. To treat this problem, in this paper, an intelligent blending strategy without needing segmentation masks is proposed. It is based on the fact that for each point in the generated sprite, the corresponding pixels in most frames belong to background and only few belong to moving objects. A counting schema is provided to make only background points participate in average blending. The experimental result shows that the visual quality of the generated sprite using the proposed blending strategy is close to that using manually segmented masks and is better than that generated by Lu-Gao-Wu method. No ghostlike shadows are produced. Furthermore, a uniform feature point extraction method is proposed to increase the precision of global motion estimation, the effectiveness of this part is presented by showing the comparison results with other existing method.

Automatic face recognition is becoming increasingly important due to the security applications derived from it. Although the facial recognition problem has focused on 2D images, recently, due to the proliferation of 3D scanning hardware, 3D face recognition has become a feasible application. This 3D approach does not need any color information. In this way, it has the following main advantages in comparison to more traditional 2D approaches: (1) being robust under lighting variations and (2) providing more relevant information. In this paper we present a new 3D facial model based on the curvature properties of the surface. Our system is able to detect the subset of the characteristics of the face with higher discrimination power from a large set. The robustness of the model is tested by comparing recognition rates using both controlled and noncontrolled environments regarding facial expressions and facial rotations. The difference between the recognition rates of the two environments of only 5% proves that the model has a high degree of robustness against pose and facial expressions. We consider that this robustness is enough to implement facial recognition applications, which can achieve up to 91% correct recognition rate. A publish 3D face database containing face rotations and expressions has been created to achieve the recognition experiments.

An approach to integrating the global and local kernel-based automated analysis of vocal fold images aiming to categorize laryngeal diseases is presented in this paper. The problem is treated as an image analysis and recognition task. A committee of support vector machines is employed for performing the categorization of vocal fold images into healthy, diffuse and nodular classes. Analysis of image color distribution, Gabor filtering, cooccurrence matrices, analysis of color edges, image segmentation into homogeneous regions from the image color, texture and geometry view point, analysis of the soft membership of the regions in the decision classes, the kernel principal components based feature extraction are the techniques employed for the global and local analysis of laryngeal images. Bearing in mind the high similarity of the decision classes, the correct classification rate of over 94% obtained when testing the system on 785 vocal fold images is rather encouraging.

This paper deals with color image simplification using levelings. This class of connected filters suppresses details but preserves the contours of the remaining structures or objects. As the notion of "color structure" is not trivial, the formulation of morphological operators for color images involves many open issues. The principle choice of a well-defined color space is crucial and it is proposed to work on a luminance/saturation/hue representation defined by the norm L₁. A family of morphological color operators is then introduced using the classical formulation with total orderings by means of lexicographic cascades. In this framework, a methodology for color image simplification is introduced, which takes advantage of a saturation-controlled combination of the chromatic and the achromatic (or the spectral and the spatio-geometric) components. More precisely, it is based on the application of a color leveling to each significant region, specifically adapted to the nature (chromatic/achromatic) of the region and which needs an initial image partition into the homogenous regions. Experimental results illustrate the performance of the new developed algorithms.

When a monocular vision-based unmanned aerial vehicle (UAV) is flown to the final approach to intercept the glide slope, the position and orientation of the airport runway in the image must be detected accurately for a host of suitable procedures to be followed. The approaching marking on the runway is showed as some white spots of high intensity as well as the complicated backgrounds. In our paper, we use pin-hole perspective principle, the constraint condition of the rectangle in inertial space, the front shot constraint condition of the target, as well as the clustering algorithm to identify the runway and output its position and orientation in image space. The results of the experiments show that by this algorithm, even from a place far away from the runway with marks being unclear, effective detection is possible. After all, single-frame detection errors exist, so we extend the basic runway-detection algorithm to the runway tracking. A full filtering strategy using particle filter can guard against potentially catastrophic results and improve the detection rate. Apparently, the whole algorithm of our paper can be treated as a special vision sensor for landing equipment of UAV.

Linear Discriminant Analysis (LDA) is a popular feature extraction technique in statistical pattern recognition. However, it often suffers from the small sample size problem when dealing with high-dimensional data. Moreover, while LDA is guaranteed to find the best directions when each class has a Gaussian density with a common covariance matrix, it can fail if the class densities are more general. In this paper, a novel nonparametric linear feature extraction method, nearest neighbor discriminant analysis (NNDA), is proposed from the view of the nearest neighbor classification. NNDA finds the important discriminant directions without assuming the class densities belong to any particular parametric family. It does not depend on the nonsingularity of the within-class scatter matrix either. Then we give an approximate approach to optimize NNDA and an extension to k-NN. We apply NNDA to the simulated data and real world data, the results demonstrate that NNDA outperforms the existing variant LDA methods.

Long range or multistep-ahead time series forecasting is an important issue in various fields of business, science and technology. In this paper, we have proposed a modified nearest neighbor based algorithm that can be used for long range time series forecasting. In the original time series, optimal selection of embedding dimension that can unfold the dynamics of the system is improved by using upsampling of the time series. Zeroth order cross-correlation and Euclidian distance criterion are used to select the nearest neighbor from up-sampled time series. Embedding dimension size and number of candidate vectors for nearest neighbor selection play an important role in forecasting. The size of embedding is optimized by using auto-correlation function (ACF) plot of the time series. It is observed that proposed algorithm outperforms the standard nearest neighbor algorithm. The cross-correlation based criteria shows better performance than Euclidean distance criteria.