Skin detection has gained popularity and importance in the computer vision community. It is an essential step for important vision tasks such as the detection, tracking and recognition of face, segmentation of hand for gesture analysis, person identification, as well as video surveillance and filtering of objectionable web images. All these applications are based on the assumption that the regions of the human skin are already located. In the recent past, numerous techniques for skin colour modeling and recognition have been proposed. The aims of this paper are to compile the published pixel-based skin colour detection techniques to describe their key concepts and try to find out and summarize their advantages, disadvantages and characteristic features.
Skin colour is an important visual cue for face detection, face recogmtlon, hand segmentation for gesture analysis and filtering of objectionable images. In this paper, the adaptive skin color detection model is proposed, based on two bivariate normal distribution models of the skin chromatic subspace, and on image segmentation using an automatic and adaptive multi-thresholding technique. Experimental results on images presenting a wide range of variations in lighting condition and background demonstrate the efficiency of the proposed skin-segmentation algorithm.
Recently, license plate detection has been used in many applications especially in transportation systems. Many methods have been proposed in order to detect license plates, but most of them work under restricted conditions such as fixed illumination, stationary background, and high resolution images. License plate detection plays an important role in car license plate recognition systems because it affects the accuracy and processing time of the system. This work aims to build a Car License Plate Detection (CLPD) system at a lower cost of its hardware devices and with less complexity of algorithms’ design, and then compare its performance with the local CAR Plate Extraction Technology (CARPET). As Malaysian plates have special design and they differ from other international plates, this work tries to compare two likely-design methods. The images are taken using a web camera for both the systems. One of the most important contributions in this paper is that the proposed CLPD method uses Vertical Edge Detection Algorithm (VEDA) to extract the vertical edges of plates. The proposed CLPD method can work to detect the region of car license plates. The method shows the total time of processing one 352x288 image is 47.7 ms, and it meets the requirement of real time processing. Under the experiment datasets, which were taken from real scenes, 579 out of 643 images were successfully detected. Meanwhile, the average accuracy of locating car license plate was 90%. In this work, a comparison between CARPET and the proposed CLPD method for the same tested images was done in terms of detection rate and efficiency. The results indicated that the detection rate was 92% and 84% for the CLPD method and CARPET, respectively. The results also showed that the CLPD method could work using dark images to detect license plates, whereas CARPET had failed to do so.
Plurality voter is one of the commonest voting methods for decision making in highly-reliable applications in which the reliability and safety of the system is critical. To resolve the problem associated with sequential plurality voter in dealing with large number of inputs, this paper introduces a new generation of plurality voter based on parallel algorithms. Since parallel algorithms normally have high processing speed and are especially appropriate for large scale systems, they are therefore used to achieve a new parallel plurality voting algorithm by using (n/log n) processors on EREW shared-memory PRAM. The asymptotic analysis of the new proposed algorithm has demonstrated that it has a time complexity of O(log n) which is less than time complexity of sequential plurality algorithm, i.e. O (n log n).
The rapid development of roads and the increasing number of vehicles have complicated road traffic enforcement in many countries due to limited resources of the traffic police, specifically when traffic infraction registration is done manually. The efficiency of the traffic police can be improved by a computer-based method. This study focused on mobile traffic infraction registration system benchmarking which is used to evaluate the server performance under load. The study attempts to provide a clear guideline for the performance evaluation of mobile road traffic infraction registration system, whereby the traffic police can make decision based on them to migrate from the manual-method toward computer-based method. A closed form of benchmark tool was used for the evaluation of the system performance. The tool was configured to imitate ramp scenarios, and statistics were gathered. The server was monitored at different times and works. Contributing factors include bottleneck, traffic, and response time, which are related with criteria and measurements. The system resource was also monitored for the tests.
The use of biometric features, to authenticate users of different applications, is growing rapidly in recent years, according to the high sensitivity of the protected information and the good security that biometric authentication provides. In this study, a method is proposed to measure the similarity between two fingerprint images, using convolutional neural networks, instead of classifying them. Thus, modifying the users that the proposed method can recognize is a matter of adding or removing model images of the users’ fingerprints. The similarity between the fingerprint image and every model image was measured in order to select the user with the highest similarity to the input image as the recognized user, where that similarity measure was compared to a threshold value in order to authenticate that user. The evaluation results of the proposed method, using FVC2002_DB1 and FVC2004_DB1 showed that the proposed method had 99.97% accuracy with 0.035% False Acceptance Rate (FAR) and 0% False Rejection Rate (FRR). Hence, the proposed method has been able to maintain high accuracy while eliminating the vulnerabilities of biometric authentication systems imposed by the use of separate stages for features extraction and similarity measurement.