Jiatao Song

A robust eye detection method using combined binary edge and intensity information

In this paper, a new eye detection method is presented. The method consists of three steps: (1) extraction of binary edge images (BEIs) from the grayscale face image based on multi-resolution wavelet transform, (2) extraction of eye regions and segments from BEIs and (3) eye localization based on light dots and intensity information. In the paper, an improved face region extraction algorithm and a light dots detection algorithm are proposed for better eye detection performance. Also a multi-level eye detection scheme is adopted. Experimental results show that a correct eye detection rate of 98.7% can be achieved on 150 Bern images with variations in views and gaze directions and 96.6% can be achieved on 564 AR images with different facial expressions and lighting conditions.

Fingertip detection and tracking using 2D and 3D information

In this paper, a method for the tracking of fingertip in scenes with complex background is proposed. Our method uses a trinocular stereo vision system, named Digiclops, to capture images and to segment finger blocks. For the case that the finger perpendicularly points to a camera, the fingertip can be rapidly located using a statistic-based method. While for other cases, a robust algorithm for fingertip detection is presented in this paper. Experimental results show that the proposed method can accurately track the fingertip at a speed of about 9.1 frames per second for the 1024times768 images and can be used for the construction of Perceptual User Interface (PUI) system.

Wavelet-based Illumination Compensation for Face Recognition using Eigenface Method

Eigenface is an effective human face recognition technique. However its performance is affected to some extent by lighting conditions in capturing face images. In this paper, a new algorithm for the illumination compensation of face images based on Wavelet transform is proposed. By discarding the low frequency components of a face image and reconstructing the image using the high frequency components only, the influence of lightings on face image recognition can be greatly reduced. Experimental results on Yale face database and our own face database show that the recognition rate of the eigenface method on the lighting compensated images is increased significantly

Bark classification by combining grayscale and binary texture features

In this paper, a texture feature based bark classification method is presented. Our method uses two types of texture features: the co-occurrence matrix feature and the long connection length emphasis (LCLE) feature, which is extracted from the binary bark image. For the extraction of binary texture maps, an improved wavelet-based edge detection algorithm is proposed. It includes two binarization steps and a post-processing step. The paper also presents an approach to combine two feature sets. Experiments on 18 different tree species, and in total 90 bark images, show that a combination of these two feature sets can achieve a much higher bark classification rate than that when each feature set is utilized individually.

Face recognition by fusing binary edge feature and second-order mutual information

Illumination change and expression variation are two main factors affecting the performance of some existing face recognition algorithms. Edge feature based methods are robust to illumination change and are easy to implement. But they donpsilat work well for the images with expression variation. In this paper, a novel face recognition method based on the fusion of binary edge feature and grayscale information is proposed to improve both the illumination and expression robustness. The second-order mutual information (MI2) is introduced as the similarity metric of grayscale face images. Experimental results on AR dataset and Yale dataset, both with illumination and expression changes, show that the overall face recognition rate of the proposed method is better than that of some commonly used approaches, indicating that our method is more effective for practical uses.

Robust nose detection and tracking using gentleboost and improved Lucas-Kanade optical flow algorithms

The problem of face feature points detection is an important research topic in many fields such as face image analysis and human-machine interface. In this paper, we propose a robust method of 2D nose detection and tracking system. This system can be valuable for disabled people or for cases where hands are busy with other tasks. The required information is derived from video data captured with an inexpensive web camera. Position of the nose tip is determined with the use of a Gabor wavelet feature based GentleBoost detector. Once the nose tip is initially located, an improved Lucas-Kanade optical flow method is used to track the nose tip feature point. Experiments show that our system is able to process 18 frames per second at a resolution of 320×240 pixels. This method will in future be used in a non-contact interface for disabled users.

Face recognition based on binary template matching

In this paper, a novel face recognition method based on binary face edges is presented to deal with the illumination problem. The Binary Face Edge Map (BFEM) is extracted using the Locally Adaptive Threshold (LAT) algorithm. Based on BEFM, a new image similarity metric is proposed. Experimental results show that face recognition rates of 76.32% and 82.67% are achieved respectively on 798 AR images and 150 Yale images with changed lighting conditions and facial expression variations when one sample per subject is used as the target image. The proposed method takes less time for image matching and outperforms some existing face recognition approaches, especially in changed lighting conditions.

Particle Filtering Object Tracking Based on Texture and Color

Object tracking based on color feature often fails in a complex background. To deal with this problem, a particle filtering object tracking approach is proposed in this paper based on local binary pattern and color feature. Color histogram is the global description of targets in color image, while local binary pattern texture contains information of neighbor region texture in gray image. These two features may complement each other, thus target is represented by both histogram of color and local binary pattern which are combined under the frame of particle filtering. The experimental results show that the proposed method effectively improves the accuracy and robustness of tracking.

A genetic algorithm with age and sexual features

Genetic Algorithm (GA) has been successfully applied to many optimization problems. One problem with Standard GA is its premature convergence for complex multi-modal functions. To overcome it, in this paper a novel genetic algorithm with age and sexual features is proposed. Age and sexual features are provided to individuals to simulate the sexual reproduction popular in nature. During applying age and sexual operators, different evolutionary parameters are given to genetic individuals. As a result, the proposed Genetic Algorithm can combat premature convergence and maintain the diversity of population, and thereby converge on global solutions.

Expression recognition based on EPCA and SVM

Equable principal component analysis (EPCA) is a powerful technique of feature extracting. It can reduce a large set of correlated variables to a smaller number of uncorrelated components. Support vector machines (SVM) is a novel pattern classification approach. It is very efficient in solving clustering problems that are not linearly separable. This paper presents a method of expression recognition based on the EPCA and SVM. According to the EPCA extracting feature, this paper recognizes expression with SVM. The multi-class classification problem is solved by the approach of one-against all SVM classifier. Experiments of human who participates in test have been trained or not are performed on the JAFFE and Yale database. And compared to the nearest classifier, the EPCA and SVM can get better recognition ratio. Therefore, it is feasible to apply EPCA and SVM to expression recognition.