Mohd Fo'Ad Rohani

KSII Transactions on Internet and Information Systems TIIS Paper Details

Underwater Wireless Sensor Networks (UWSNs) is interesting area for researchers.To extract the information from seabed to water surface the the majority numbers of routing protocols has been introduced. The design of routing protocols faces many challenges like deployment of sensor nodes, controlling of node mobility, development of efficient route for data forwarding, prolong the battery power of the sensor nodes, and removal of void nodes from active data forwarding paths. This research article focuses the design of the Reliable Multipath Energy Efficient Routing (RMEER) which develops the efficient route between sensor nodes, and prolongs the battery life of the nodes. RMEER is a scalable and robust protocol which utilizes the powerful fixed courier nodes in order to enhance the network throughput, data delivery ratio, network lifetime and reduces the end-to-end delay. RMEER is also an energy efficient routing protocol for saving the energy level of the nodes. We have used the NS2.30 simulator with AquaSim package for performance analysis of RMEER.We observed that the simulation performance of RMEER is better than D-DBR protocol.

Skin detection is an important step in many computer vision applications. It has been employed in face detection, hand gesture recognition, illicit image filtering, steganography and content based image retrieval. This is due to the skin colour that attractive feature in detecting the skin in coloured image. In contrast, skin colour detection suffers in low accuracy due to colour properties between the real skin surface and the skin-like objects. Therefore, this paper proposes a dynamic skin colour detection using multi-scales online skin sampling approach. This dynamic skin colour detection involved two procedures for generating the dynamic threshold in colour spaces. Moreover, six colour spaces have been studied to find the best colour models for our proposed method. The first procedure is the online skin sampling that obtained directly from the face candidates to generate the dynamic threshold values of each studied colour spaces. Alongside with the first procedure, we obtained optimal scale for skin sample with 0.25, 0.2 reduction. Meanwhile, the second procedure known as skin pixel classification uses the dynamic threshold obtained from the first procedure to classify the skin in the image. We achieved a satisfactory result in term of precision, recall, accuracy and F₁. The experimental result shows that the proposed dynamic skin colour detection achieved good performance via YCr colour model.

Color space conversion utilized by many researchers in order to enhance skin recognition performance by projecting the skin color cluster to a more distinctive distribution. In spite of the substantial research effort in this area, finding a suitable color space for face and skin recognition is still an unsolved issue. Deviation of skin tone under different lighting condition, dissimilarity of skin color among different ethnics and races, various camera sensors characteristics, presence of skin-like color objects in image background and variation of skin color tone among different body limbs are among the major challenges in skin recognition. Majority of these challenges are expected to be mitigated through color space conversion. This paper proposes a new hybrid color space by applying Principal Component Analysis technique to skin color cluster in ten existing conventional color spaces including RGB, YC _b C _r , YUV, nRGB, i ₁ i ₂ i ₃ , YIQ, XYZ, YP _b P _r , YES, YC _g C _r . The proposed hybrid color space which termed P ₁ P ₂ P ₃ consist of the three major Principal Components of these conventional color spaces components. Using Algebraic simplification these principal component has been reformulated in terms of RGB color space. Parametric pixel wised skin detection techniques have been employed in order to evaluate the proposed color space effect on skin detection performance. Three popular supervised classifiers including Multilayer Perceptron, Support Vector Machine and Random Forest has been employed to generate a parametric model of skin color cluster using the proposed color space. Experiment results shows the proposed hybrid color space P ₁ P ₂ P ₃ with F-score and False Positive Rate 0.960 and 0.041 respectively performed better than the existing conventional color spaces in term of pixel wised skin recognition.

Matching features of all image overlapping blocks is a common method to explore duplicated regions. In this way, the block size should be small in order to detect small forgery regions. The problem is how to extract a robust feature for small block size against rotation and compression. In this paper, a new spatial domain feature for block size 8×8 pixels namely Spiral Unique Sequence (SUS) is proposed. The experimental results and SUS feature matching demonstrate that this feature is robust against rotation and compression.

Image block matching is the main step of duplicated region detection for exploring copy-paste image forgery. High computational time in this step is one of the most important problems to find similar regions. In this paper we propose an efficient image block matching algorithm based on two layer feature extraction in order to improve time complexity. Furthermore, we determine performance of proposed algorithm based on time complexity function. The experimental results and mathematical analysis demonstrate that two layer matching can be more time-efficient than previous common methods such as lexicographically sorting.

This paper proposes a Multi-Level Sampling (MLS) approach for continuous Loss of Self-Similarity (LoSS) detection using iterative window. The method defines LoSS based on Second Order Self-Similarity (SOSS) statistical model. The Optimization Method (OM) is used to estimate self-similarity parameter since it is fast and more accurate in comparison with other estimation methods known in the literature. Probability of LoSS detection is introduced to measure continuous LoSS detection performance. The proposed method has been tested with real Internet traffic simulation dataset. The results demonstrate that normal traces have probability of LoSS detection below the threshold at all sampling levels. Meanwhile, false positive detection can occur where abnormal traces have probability of LoSS that imitates normal behavior at sampling levels below 100ms. However, the LoSS probability exceeds the threshold at sampling levels larger than 100ms. Our results show the possibility of detecting anomaly traffic behavior based on obtaining continuous LoSS detection monitoring

This paper investigates loss of self-similarity (LoSS) detection performance using exact and asymptotic second order self-similarity (ESOSS and ASOSS) models. Previous works on LoSS detection have used ESOSS model with fixed sampling that we believe is insufficient to reveal LoSS detection efficiently. In this work, we study two variables known as sampling level and correlation lag in order to improve LoSS detection accuracy. This is important when ESOSS and ASOSS models are considered concurrently in the self-similarity parameter estimation method. We used the optimization method (OM) to estimate the self-similarity parameter value since it was proven faster and more accurate compared to known methods in the literature. Our simulation results show that normal traffic behavior is not influenced by the sampling parameter. For abnormal traffic, however, LoSS detection accuracy is very much affected by the value of sampling level and correlation lag used in the estimation.

This paper analyzes loss of self-similarity (LoSS) detection accuracy using parameterpsilas adjustment which includes different values of sampling level and correlation lag. This is important when considering exact and asymptotic self-similar models concurrently in the self-similarity parameter estimation method. Due to the needs of high accuracy and fast estimation, the optimization method (OM) based on second order self-similarity (SOSS) statistical model was proposed in the previous works to estimate self-similarity parameter. Consequently, curve fitting error (CFE) value estimated from OM is used to detect LoSS efficiently. This work investigates the effect of the parameterpsilas adjustment for improving the CFE accuracy and estimation time speed. We have tested the method with real Internet traffics simulation that consists of normal and malicious packets traffic. Our simulation results show that LoSS detection accuracy and estimation time can be affected by the chosen of sampling level and correlation lag values.