1 2538-4201 Research Center on Developing Advanced Technologies 181 Paper Improving the nonlinear manifold separator model to the face recognition by a single image of per person Seyyedsalehi Seyyede Zohreh Seyyedsalehi Seyyed Ali 1 6 2015 12 1 3 16 17 11 2013 21 04 2015 Manifold learning is a dimension reduction method for extracting nonlinear structures of high-dimensional data. Many methods have been introduced for this purpose. Most of these methods usually extract a global manifold for data. However, in many real-world problems, there is not only one global manifold, but also additional information about the objects is shared by a large number of manifolds. In this context, based on previous researches, this paper proposes a nonlinear dimension reduction method based on the deep neural network that extract simultaneously manifolds embedded in data. In nonlinear manifold separator model, unlike unsupervised learning of bottleneck neural network, data labels are indirectly used for manifold learning. Given the deep structure of the model, it has been shown that using pre-training methods can significantly improve its performance moreover, to improve within-manifold discrimination for different classes, its standard functions have been improved. This paper makes use of the model for extracting both expression and identity manifolds for facial images of the CK+ database. In comparing early and improved models, it is shown that the facial expression recognition rate from 24.29% to 75.07% and the face recognition rate by a single image of each person by enriching dataset from 90.62% to 97.07% were improved.
64 Paper IMS SIP Server security model using the TVRA methodology Asgharian Hassan taj nasrin 1 6 2015 12 1 17 32 03 06 2013 24 02 2015 IMS (IP Multimedia Subsystem) network is considered as an NGN (Next Generation Network) core networks by ETSI. Decomposition of IMS core network has resulted in a rapid increase of control and signaling message that makes security a required capability for IMS commercialization. The control messages are transmitted using SIP (Session Initiation Protocol) which is an application layer protocol. IMS networks are more secure than typical networks like VoIP according to mandatory of user authentication in registration time and added SIP signaling headers. Also different vulnerabilities have been occurred that lead to SIP servers attacks. This paper studies the main SIP servers of IMS (x-CSCF) based on ETSI Threat, Vulnerability and Risk Analysis (TVRA) method. This method is used as a tool to identify potential risks to a system based upon the likelihood of an attack and the impact that such an attack would have on the system. After identifying the assets and weaknesses of IMS SIP servers and finding out the vulnerabilities of these hardware and software components, some security hints that can be used for secure deployment of IMS SIP servers are proposed. Modeling shows the effects of server weaknesses and threats that reduces availability. Any designed system has some assets with weaknesses. When threats have accrued based on weaknesses, the system will vulnerable. Vulnerability analysis optimizes costs and improves security. 165 Paper A Feature-based Vehicle Tracking Algorithm Using Merge and Split-based Hierarchical Grouping Pourghassem Hossein g g Najafabad Branch, Islamic Azad University 1 6 2015 12 1 33 46 12 09 2013 18 10 2014 Vehicle tracking is an important issue in Intelligence Transportation Systems (ITS) to estimate the location of vehicle in the next frame. In this paper, a feature-based vehicle tracking algorithm using Kanade-Lucas-Tomasi (KLT) feature tracker is developed. In this algorithm, a merge and split-based hierarchical two-stage grouping algorithm is proposed to represent vehicles from the tracked features. In the proposed grouping algorithm, with defining measures such as distance, spread and also blob analysis, initial grouping results formed by K-means clustering algorithm are refined. Moreover, to modify the performance of KLT tracker and also optimized utilization from grouping results obtained by proposed algorithm, an effective group matching algorithm based on a merging and splitting scheme is employed to match the tracked groups from a frame to the next frame. The proposed tracking algorithm is evaluated on different test videos with various illumination conditions such as day, night and shadow. The obtained results show that our proposed tracking algorithm covers the most challenges of tracking in the ITS applications. 120 Paper Iranian License Plate identification with fuzzy support vector machine Montazer Gholam Ali shayestehfar mohammad 1 6 2015 12 1 47 56 12 06 2013 18 04 2015 License plate recognition is one of the most important applications used in intelligent transportation systems. Difficulty of correct detection and identification of the car plates in different environment conditions makes researchers try new approaches to better solve the problem. License plate recognition problem is divided into three sub problems: "Plate Location", "Character Segmentation", and "Character Identification". In this paper we have tried to improve location and identification of Iranian license plate with fuzzy rules. License locating has been done with edge detection, morphological operations and using fuzzy rules and characters have been identified by fuzzy support vector machine. By applying the algorithm on 50 images, 90% of plates were located and 94% of characters were identified successfully. This shows superiority of our algorithm over non-fuzzy approaches. 43 Paper Robustness of Motion Vector against Channel Error for Improvement of Synthesized Video Quality Farsi Hassan Etezadifar Pouria 1 6 2015 12 1 57 78 20 05 2013 11 03 2015 According to progress of technology during the recent decades, video transmission through a wireless channel has found high demands. In this field, several methods have been proposed to improve video quality. Appearing error in motion vector values is one of the most important factors which can affect the video quality. In case of creating errors in motion vector, the synthesized video frames are moved compared to the previous situation and therefore the synthesized video quality considerably degrades. In this paper, in order to overcome this problem and also to increase PSNR, we propose a method to increase the channel coding rate but transmission rate is maintained constant. In the proposed method, firstly, the motion vector is searched in each block with size of 8*8. After ending the search, the adjacent blocks with the motion vector equals to zero (without movement) are combined together and provide bigger block. Meanwhile, the blocks with equal motion vectors are combined together and transmitted to receiver in two different methods. The experimental results show that the proposed method without increasing side information is able to provide more robustness for video frame against channel errors. The performance of the proposed method has been compared with the new method for different source coding rates and SNRs. 180 Paper Prediction of consonants Intelligibility for Listeners with Normal Hearing Using Microscopic Models of Speech Perception Considering Different Distance Measures in Automatic Speech Recognizer Geravanchizadeh Masoud l Fallah Ali m Eterafoskouei Mirali n l University of Tabriz m University of Tabriz n Tabriz University of Medical Science 1 6 2015 12 1 79 90 16 11 2013 20 04 2015 In this study, recognition rates of consonants available in vowel-consonant-vowel structure in hearing tests and two microscopic models will be investigated. Such a syllable structure doesn’t exist in Farsi and Azerbaijani languages, but since the goal is only recognition of middle phoneme, according to hearing tests, listeners are able to properly recognize phonemes in clean speech conditions. Inasmuch as these syllable structures are meaningless, it will be suitable for our purpose that is only determination of recognition rates of phonemes not meaningful words. Using this corpus, listeners’ linguistic knowledge in prediction of words is disregarded. Results of hearing tests are compared with two microscopic models based on human auditory system. Difference between two models is at the final stage of feature extraction that in first model, a 8 Hz filter and in the second model a modulation filterbank is used. Correct recognition rates of phonemes in different signal to noise ratios and two distance metrics for speech recognizer, will be compared. In this study recognition rates of consonants for listeners with Azerbaijani native language have been studied. Beside the empirical aspect of the paper, the innovations of this work lies in the study of using two different distance measures for Holube’s model and also direct comparison of two microscopic models in prediction of overall recognition rates and recognition rate of each consonant.