Puteh Saad

A comparison of circular object detection using Hough transform and chord intersection

In this paper, the circular Hough transform (CHT) and the chord intersection have been used to find the circular object in the feature extraction process. The chord intersection technique does not require any gradient information which may be sensitive to noise meanwhile for the CHT technique, the gradient information has been used. In this research, the coconut was selected as the object of interest. 40 images have been experimented to evaluate the performance of the techniques and the detection rate for the CHT is 92.5% and 85% for the chord intersection technique. The average computational time for chord intersection technique is 0.1495s by CPU (AMD Athlon 64x2 Dual core 3800) 2GHz meanwhile CHT consumed more time, 2.3871s in detecting the circular pattern.

Multiple neural networks for Human Face Recognition

In this paper, a computational model has been developed to identify a face of an unknown person’s by utilizing eigenfaces as unique features and backpropagation Neural Network for recognition. The features of a basic human face are extracted using eigenfaces. These features are then used to identify an unknown face by using multiple numbers of backpropagation neural networks. Samples of 15 human faces are obtained from The ORL database. The experiments are compared to the effects of changes size of face images, different face images combination and different neural network parameter. The classification more than 90% for trained classes and 18% for untrained classes were achieved.

Classification of EEG signals for brain-computer interface applications: Performance comparison

This paper presents a comparison of Electroencephalogram (EEG) signals classification for Brain Computer-Interfaces (BCI). At present, it is a challenging task to extract the meaningful EEG signal patterns from a large volume of poor quality data and simultaneously with the presence of artifacts noises. Selection of the effective classification technique of the EEG signals at classification stage is very important to get the robust BCI system. Support Vector Machine (SVM), k-Nearest Neighbour (k-NN), Multi-Layer Perceptron Artificial Neural Network (MLP-ANN) and Logistic Regression (LR) were evaluated in this paper. A BCI competition IV — Dataset 1 is used for testing the classifiers. It is shown that LR and SVM are the most efficient classifier with the highest accuracy of 73.03% and 68.97%.

Investigation of information fusion in face and palmprint multimodal biometrics

This paper reviews several information fusion techniques and strategies in the application of multimodal biometrics system using face and palmprint images. Multimodal biometric is able to overcome several limitations in single modal biometric such as intra-class variations, less discriminative power, noise data and redundant features. By consolidating two kinds of modality a better performance can be achieved. Information fusion in multimodal biometrics can be carried out at three possible levels, i.e. feature, matching score and decision levels. Fusions at these three levels have their own attributes, thus this paper is aimed to compare their effectiveness. A specific fusion rule is necessary to combine the information at each level. Several numbers of analyses on verification and identification shows matching score fusion is able to achieve the best performance which is 98% recognition rates and 98.5% GAR at 0.1% FAR when tested using AR face and PolyU palmprint datasets.

Hyper-threading technology: Not a good choice for speeding up CPU-bound code

Hyper-threading (HT) technology allows one thread to execute its task while another thread is stalled waiting for shared resource or other operations to complete. Thus, this reduces the idle time of a processor. If HT is enabled, an operating system would see two logical cores per each physical core. This gives one physical core the ability to run two threads simultaneously. However, it does not necessarily speed up the performance of a parallel code twice the number of physical cores. This happens when two threads are trying to access the shared CPU resource. The instructions could only be executed one after another at any given time. In this case, parallel CPU-bound code could attain a little improvement in terms of speedup from HT on a quad-core platform, which is Intel i5-2410M@2.30GHz.

A Review on DSTATCOM Neural Network Control Algorithm for Power Quality Improvement

This paper reviews neural network control algorithm for power quality improvement. Further, this paper focuses on the neural network control algorithm for DSTATCOM and surveys its area of improvements. Various architectures of Neural Network such as Adaline/Widrow-Hoff, perceptron, Back-propagation (BP), Hopfield, and Radial Basis Function (RBF) that has been reviewed in this paper. It is found that many researches on theoretical works and single phase system are widely performed, whereas its application on distribution network for three phase system is hardly found. Even so much improvement that have been done by researchers theoretically to improve the drawbacks of Neural Network controller; there are still wide gaps for verification through experimental implementation and industrial applications.

Improving hybrid speaker verification in noisy environments using least mean-square adaptive filters

In this paper, we present a hybrid speaker verification system based on the Hidden Markov Models (HMMs) and Vector Quantization(VQ) and Least Mean-Square (LMS) adaptive filtering. The aim of using hybrid speaker verification is to improve the HMMs performance, while LMS adaptive filtering is to improve the hybrid speaker verification performance in noisy environments. A Malay spoken digit database is used for the training and testing. It is shown that, in a clean environment a Total Success Rate (TSR) of 99.97% is achieved using hybrid VQ and HMMs. For speaker verification, the true speaker rejection rate is 0.06% while the impostor acceptance rate is 0.03% and the equal error rate (EER) is 11.72%. In noisy environments without LMS adaptive filtering TSRs of between 62.57%-76.80% are achieved for Signal to Noise Ratio (SNR) of 0-30 dBs. Meanwhile, after LMS filtering, TSRs of between 77.31%-76.87% are achieved for SNRs of 0-30 dB.