Home
Editorial Board
Aims & Scopes
Author's Guide
Review Policy
Paper Submission
Special Issue
Contact

Vol. 5, No. 1:

Entropic Image Segmentation: A Fuzzy Approach Based on Tsallis Entropy

Samy Sadek, Sohag University, Egypt
Ayoub Al-Hamadi, Otto-von-Guericke-University Magdeburg, Germany
Abstract:
In this paper, a fuzzy approach for image segmentation based on Tsallis entropy is introduced. This approach employs fuzzy Tsallis entropy to measure the structural information of image and to locate the optimal threshold desired by segmentation. The proposed method draws upon the postulation that the optimal threshold concurs with maximum in- formation content of the distribution. The contributions of the work are as follow: Initially, fuzzy Tsallis entropy as a measure of spatial struc- ture of image is described. Then, an unsupervised entropic segmentation method based on fuzzy Tsallis entropy is developed. Although the proposed approach belongs to entropic segmentation approaches (i.e., such approaches are commonly applied to grayscale images), it is adapted to be viable for segmenting color images. Finally, substantial experiments are carried out on realistic images to validate the effectiveness, efficiency and robustness of the proposed method.
Full Paper (in PDF)

Computer Vision based Vehicle Recognition on Indian Roads

R.S Vaddi, L.N.P Boggavarapu, K.R Anne, V.R Siddhartha Engineering College, Vijayawada, India
Abstract:
Feature extraction and classification are two most important modules for any vision-based object recognition system. In the case of vehicles, most of the methods in these modules found to be less accurate in recognition even though they work well for other objects. We are interested in recognition of vehicles on Indian roads. There are number of challenges in implementing vehicle recognition in Indian scenario like bad road conditions, traffic rules violation and variance among vehicles, etc. In order to overcome these difficulties, we implemented feature extraction module using bag of features (combination of Harris-corner detector and SIFT features), and classification is performed using Support Vector Machines (SVM). To validate our proposed method, we have introduced Indian Vehicle Database. The images in this database are extracted from daylight Indian urban traffic scenes. Our proposed method achieves 40-45 percent improvement over the baseline methods.
Full Paper (in PDF)

Enhanced Contour Description for Pedestrian Detection

Xiaoyun Du, Robert Laganiere, Si Wu, University of Ottawa, Canada
South China University of Technology, China
Abstract:
In this paper, an enhanced contour descriptor is presented for pedestrian detection, an important and challenging task in computer vision. A discriminative feature is essential to build a robust detection. Histograms of Oriented Gradients (HOG) and Local Binary Patterns (LBP) are well known descriptors with proven efficiency. We propose here the use of Variational LBP (VLBP) as an auxiliary feature and combine it with HOG to generate a discriminative contour descriptor used in our detection model. We also perform feature selection by using the Feature Generating Machine in order to select the most useful elements of a descriptor vector without impacting on the classiffication performance. Moreover, a two-layer cascade model is proposed to achieve both accurate detection and lower computational complexity. The intersection kernel based support vector machine is employed in our cascade model as a performing classi cation tool that integrates well histogrammic features. A boot- strapping algorithm is also applied in our training procedure to improve the performance of classi cation. The result of the experiment shows that our approach achieves good performances on the INRIA benchmark dataset. A conclusion of this work is that an enhanced and concise descriptor that combines HOG and VLBP can improve contour description and thus leads to better detection performance.
Full Paper (in PDF)

Personnel Recognition in the Military using Multiple Features

Martins E. Irhebhude, Eran A. Edirisinghe Loughborough University, UK
Abstract:
This paper presents an automatic, machine vision based, military personnel identification and classification system. Classi cation was done using a Support Vector Machine (SVM) on sets of Army, Air Force and Navy camou age uniform personnel datasets retrieved from google images and selected military websites. In the proposed system, the arm of service of personnel is recognised by the camou age and the type of cap badge/logo of a persons uniform. The detailed analysis done include; camou age and plain cap differentiation using Gray Level Co-occurrence Matrix (GLCM) texture features; Army, Air Force and Navy camou aged uniforms differentiation using GLCM texture and colour histogram bin features; plain cap differentiation using Speed Up Robust Feature (SURF) on the cap badge. Correlation-based Feature Selection (CFS) was used to improve recognition by selecting discriminating features, thereby speeding the classification process. With this method success rates recorded during the analysis include 94% for camou age appearance category, 100%, 90% and 100% rates of plain and camou age cap categories for Army, Air Force and Navy respectively. Similarly, using SURF features on the cap badge in the top region of the segmented human part of top and bottom; the plain cap badge of the military personnel was accurately categorised. By this, we have shown that the proposed method can be integrated into a face recognition system, which recognises an individual and determine the arm of service the person belongs. Such a system can be used to enhance the security of a military base or facility. Substantial analysis has been carried out and results after comparison with two other techniques prove that the proposed method can correctly classify military personnel into various arms of service. Accurate recognition was recorded with the proposed technique.
Full Paper (in PDF)

Visual Tracking on Riemannian Space Using Updated Standard Deviation Based Model

Anuja Kumar Acharya, Biswajit Sahoo KIIT University, India
Abstract:
Object tracking using appearance based modeling from non stationary camera is one of the key aspect of visual tracking. While most of the existing algorithms are able to track objects well in controlled environ- ments, those methods usually fail to track a longer sequence of trajectory in the presence of signi cant variation of the objects appearance or sur- rounding illumination. In this paper, we propose a new simple standard deviation based model updated method for tracking a longer sequence of trajectory for the target object. Non singular covariance based fea- ture subspace is constructed for each candidate image region that lie on riemannian space. This feature subspace is updated by adding the vec- tor mean di erence of standard deviation between the referenced object and the detected objects, to each observation vector of the referenced model. The resultant covariance structure of this updated target refer- ence model can be used for tracking the next sequence video frame. In the proposed model, also we use the kalman ltering for e ectively han- dle the background clutter and temporary occlusion. Simulation result shows the current method is robust for real time tracking.
Full Paper (in PDF)

Hit Counter
Hit Counter (since 1 Nov. 2012)