No abstract

The rate of annual road accidents attributed to drowsy driving are significantly high. Due to this, researchers have proposed several methods aimed at detecting drivers' drowsiness. These methods include subjective, physiological, behavioral, vehicle-based, and hybrid methods. However, recent reports on road safety are still indicating drowsy driving as a major cause of road accidents. This is plausible because the current driver drowsiness detection (DDD) solutions are either intrusive or expensive, thus hindering their ubiquitous nature. This research serves to bridge this gap by providing a test-bed for achieving a non-intrusive and low-cost DDD solution. A behavioral DDD solution is proposed based on tracking the face and eye state of the driver. The aim is to make this research an inception to DDD pervasiveness. To achieve this, National Tsing Hua University (NTHU) Computer Vision Lab's driver drowsiness detection video dataset was utilized. Several video and image processing operations were performed on the videos so as to detect the drivers' eye state. From the eye states, three important drowsiness features were extracted: percentage of eyelid closure (PERCLOS), blink frequency (BF), and Maximum Closure Duration (MCD) of the eyes. These features were then fed as inputs into several machine learning models for drowsiness classification. Models from the K-nearest Neighbors (KNN), Support Vector Machine (SVM), Logistic Regression, and Artificial Neural Networks (ANN) machine learning algorithms were experimented. These models were evaluated by calculating their accuracy, sensitivity, specificity, miss rate, and false alarm rate values. Although these five metrics were evaluated, the focus was more on getting optimal accuracies and miss rates. The result shows that the best models were a KNN model when k = 31 and an ANN model that used an Adadelta optimizer with 3 hidden layer network of 3, 27, and 9 neurons respective. The KNN model obtained an accuracy of 72.25% with a miss rate of 16.67%, while the ANN model obtained 71.61% and 14.44% accuracy and miss rate respectively.

No abstract

Fast Fourier Transform has long been established as an essential tool in signal processing. To address the computational issues while helping the analysis work for multi-dimensional signals in image processing, sparse Fast Fourier Transform model is reviewed here when applied in different applications such as lithography optimization, cancer detection, evolutionary arts and wasterwater treatment. As the demand for higher dimensional signals in various applications especially multimedia appplications, the need for sparse Fast Fourier Transform grows higher.

Road users make vital decisions to safely maneuver their vehicles based on the road markers, which need to be correctly classified. The road markers classification is significantly important especially for the autonomous car technology. The current problems of extensive processing time and relatively lower average accuracy when classifying up to five types of road markers are addressed in this paper. Two novel real time video processing methods are proposed by extracting two formulated features namely the contour number, , and angle, 𝜃 to classify the road markers. Initially, the camera position is calibrated to obtain the best Field of View (FOV) for identifying a customized Region of Interest (ROI). An adaptive smoothing algorithm is performed on the ROI before the contours of the road markers and the corresponding two features are determined. It is observed that the achievable accuracy of the proposed methods at several non-urban road scenarios is approximately 96% and the processing time per frame is significantly reduced when the video resolution increases as compared to that of the existing approach.