- A Novel Approach for Complex Image Processing Algorithm Design and Optimisation
Moira I. Smith and Duncan L. Hickman, Tektonex Ltd, UK
ABSTRACTThe use of modelling/simulation to predict the overall performance of image processing algorithms on camera outputs or as part of an electro-optic (EO) device/system is well-established. Performance prediction of algorithms is an essential component of any development lifecycle, from requirements definition through design and test to verification, validation and acceptance. Typically prediction relies upon parametric modelling approaches which, although theoretically sound or based on trials results, suffer from limited envelopes of validity and inaccuracy because of their transfer-function/statistical approach. Simulations offer higher accuracy but much longer run times, or else less flexible real-time in-the-loop black box testing.
A novel and successful approach to this problem is presented in this paper which combines the processing speed attributes of parametric models with the accuracy of ‘image-flow’ representations in a statistically valid framework. The paper details the approach and provides an example for automatic target recognition (ATR) algorithm performance prediction and optimisation.
- Energy-Efficient Subcarrier Assignment in Uplink OFDMA Systems under QoS Constraints
Xunan Li, Chen Chen and Ye Jin , Peking University, China
ABSTRACTEnergy efficient design in uplink communication networks is attracting more and more attention due to the limited battery capacity. In this paper, we consider the uplink OFDMA networks with quality of service (QoS) constraint. We analyze the inherent relation between energy efficiency (EE) and the number of subcarriers of a user and then give the mathematical expressions to calculate the EE gain on a subcarrier. Base on the mathematical expressions, we propose a low complexity subcarrier assignment algorithm. In this algorithm, we first implement a suboptimal subcarrier assignment and then adjust the assignment in an iterative method to increase the system EE.Simulation results show that the proposed algorithm improves the energy efficiency in a low complexity compared with the exist iterative algorithm.
- Collaborative Relay Beamforming based on Minimum Power for Multiuser in Multicell Systems
Yuanzhi He1, Xianghui Hu1, Xiaoning Zhang2 and Chen Chen2, 1Institute of China Electronic System Engineering Corporation, China & 2Peking University, Beijing, China.
ABSTRACTIn this paper, we study the collaborative relay beamforming with the coordinated multicell technology to improve the cell-edge user performance in multicell systems. In particular, a novel iterative scheme is proposed to jointly optimize the base station (BS) and relay beamforming weights under the receive signal-to-interference-and-noise ratio (SINR) constraints in the perfect channel state information (CSI), while minimizing the BS and relay power for future green communications.Moreover, we prove that the optimization problems for BS and relay beamforming weights can be effectively solved by using the semidefinite relaxation (SDR) approach. Simulation results validate our theoretical analysis and demonstrate that the proposed iterative scheme can achieve near-optimal performance within a few iterations.
- Cross Correlation Analysis of Multi-channel Near Infrared Spectroscopy
Raul Fernandez Rojas1, Xu Huang1, Keng Liang Ou2 and Jehu Lopez-Aparicio3,1University of Canberra, Australia ,2Taipei Medical University, Taiwan and 3National Autonomous University of Mexico, Mexico
ABSTRACTIn this paper we present the use of a signal processing technique to find dominant channels in near infrared spectroscopy (NIRS). Cross correlation is computed to compare measuring channels to identify delays among the channels. In addition, visual inspection was used to identify potential dominant channels. The results showed that the visual analysis exposed pain-related activations in the primary somatosensory cortex (S1) after stimulation which is consistent with similar studies and the cross correlation analysis found dominant channels on both cerebral hemispheres. The analysis also showed a relationship between dominant channels and neighbouring channels. Therefore, our results present a new method to detect dominant regions in the cerebral cortex using near-infrared spectroscopy. These results have also implications in the reduction of number of channels by eliminating irrelevant channels for the experiment.
- A Fast Search Algorithm for Large Video Database Using Hog Based Features
Qiu Chen1, Koji Kotani2 , Feifei Lee3 and Tadahiro Ohmi2 , 1Kogakuin University, Japan , 2Tohoku University, Japan and 3University of Shanghai for Science and Technology, China
ABSTRACTIn this paper, we propose a novel fast video search algorithm for large video database. Histogram of Oriented Gradients (HOG) has been reported which can be reliably applied to object detection, especially pedestrian detection. We use HOG based features as a feature vector of a frame image in this study. Combined with active search, a temporal pruning algorithm, fast and robust video search can be achieved. The proposed search algorithm has been evaluated by 6 hours of video to search for given 200 video clips which each length is 15 seconds. Experimental results show the proposed algorithm can detect the similar video clip more accurately and robust against Gaussian noise than conventional fast video search algorithm.
- Image Understanding Using Object Identification and Spatial Relationship
Hamid Sadeq Mahdi , University of Diyala / Basic Education College , Iraq
ABSTRACTIU involves psycho-thinking, math and geometrical experiences/ability; it may require analyzing, sorting, classifying / categorizing/ for identification needs, including comparing for difference or gaps appreciating in an image/photo/picture or object (shape, color, appearance…, etc.) diagnosis. Such an investigations instance makes uses of various methods. IU is in background support to multiple fields of knowledge (e.g. AI, Robotics, Computer graphics and multimedia, Psychology, Image processing and objects recognition; etc.). In other hands, spatial relationship information holds huge and great inputs for high-level image understanding study. This articles review has focused on IU processes using object identification and object Spatial relationship, while remaining closer to related topics.
- Result Analysis of Facial Expression on the basis of Lip Movement for User Authentication
Dolly Shah and Upendra Dwivedi , Shri Vaishnav Institute of Technology and Science , India
ABSTRACTWe proposed an algorithm which is linked with the code of lips movement which mainly consists of that type of code which is inserted inside the motion of lip which detects the lips movement of opening and closing. This method is translate and rotation invariant which works in efficient manner which does not require lips segmentation. We provide an automatic tool which converts the video into sequence of frame and then calculate the threshold value on the basis of frame. This algorithm also removes the unwanted noise and tries to give almost accurate result. We implemented an automatic tool which gives the adaptive background and binarized image of the sequence frame. It is one of the confirmation method for double security to the visual speaker that the speaker is actually validated through individual code and also shows biometric linked with lips movement. For this we have implemented a framework which is based on lip security for this we have use many strategies like talk investigation and outward examination. Our trial effects show how proposed technique advantageously work and best in class methodology.
- Satisfying Customer Requirements by Applying Image Processing
Suchada Rianmora, Molticha Rangsiyangkoon and Sirasith Cherdboonmuang , Thammasat University , Thailand
ABSTRACTThis project focuses on the application of generating 3D model from 2D multi-views images. Only 2D visual could not be delivered a clear explanation of the requirement so non-scanning data acquisition method has been applied in this research for increasing capability of the communication between customer and manufacturer. The proposed approach can be divided into four main sections. Preparing the object for capturing purpose is started first. Then, taking images of the different views of the desired object is applied. The third step is constructing a 3D model from the 2D images. After generating 3D model, the surface reconstruction and correction is performed to enhance the quality of the 3D result. The main advantage of generating 3D model from 2D multi-views images is to capture the desired object directly and easily where the basic or compact digital camera can be applied with less skill levels required.
- Comparison of Different Digital Filter Techniques for Denoising of ECG Signal
Babak Yazdanpanah, K.Sravan Kumar and Gilda Farajollahzadeh , Andhra University, India
ABSTRACTHeart waveforms resemble arterial blood pressure, intracranial pressure and pulse oximetry, which provide useful data about respiratory rate, heart rate and pulse pressure variation. ECG is a graph which records heart waveforms which is utilized for medical diagnosis such as arrhythmia detection and myocardial infarction. This is generally impaired by various kinds of artifact like baseline wander, power line interference, high frequency noise, physiological artifacts etc. These should be removed before diagnosis. Basically, there are two types of methods which are deliberated in this paper; Extended Kalman Filters (EKF) with real ECG signal and Adaptive filters which are based on Least Mean Square (LMS) algorithm and Normalized Least Mean Square (NLMS) algorithm. Both of which propose to reduce artifacts from ECG signals. The structure and the coefficients of the digital filters are designed using an FDA tool in MATLAB ®2014a. Ultimately, we have used EKF with real ECG signals from the MIT-BIH arrhythmia database and assessed the efficiency of the conventional Adaptive filters which again are based on LMS, NLMS algorithms. In addition, performance parameters such as power spectrum density, Signal to Noise Ratio (SNR) and Mean Square Error (MSE) of both noised and denoised ECG signals are elucidated in this paper. The resulting outcome has shown that the SNR, MSE values of ECG with NLMS algorithm which has a step size of μ =0.01 is 14.3692, 0.0874 and of μ=1 is 14.3613, 0.0875 proves that NLMS provides better efficiency when compared to EKF and Adaptive filters with LMS algorithm
- Comparison between Face Detection Techniques
Mai Farag , Ahmed H. Asad , Tarek Elgazaly and Hesham A. Hefny , Cairo University , Egypt
ABSTRACTFace detection is a required first step in any face recognition system. The main objective of the face detection is to determine whether human faces exist in an image or not. If exists, it returns their spatial locations and extents in the image. Various methods were proposed to detect faces in different scales, orientations, facial expressions and illumination conditions in the recent years. The purpose of this paper is to present a comparative study of the three most popular face detection algorithms on two collected face images groups from the Facebook website, regarding their average sensitivity, specificity and accuracy ratios in two detection cases: single face (400 images) and multiple faces (300 images). These algorithms are Viola-Jones, histograms of oriented gradients and neural-network based face detectors. This comparative study indicates that the histograms of oriented gradients algorithm gives the best performance over the other two algorithms in the two detection cases.