Mochamad Hariadi

3D reconstruction of under water coral reef images using low cost multi-view cameras

This research describes a 3D reconstruction method of coral reefs using low-cost underwater cameras. We employed a multi-view camera system consisting of 3 identical waterproof cameras arrayed on a stereo-base, and collected footage of the seafloor in linear transects. To develop a 3D-representation of the seafloor image-pairs were first extracted from the video footage manually. Then corresponding points are automatically extracted from the stereo-pairs by the well known SIFT algorithm, which is invariant to scale, translation, and rotation. Based on the resultant x,y,z point cloud the 3D appearance of the coral reef is approximated by a Triangulation technique utilizing Delaunay Triangulation. The experimental result demonstrate robust 3D reconstruction with manual adjustment of camera A, B or C selection.

Osteoarthritis Classification Using Self Organizing Map Based on Gabor Kernel and Contrast-Limited Adaptive Histogram Equalization

Localization is the first step in osteoarthritis (OA) classification. Manual classification, however, is time-consuming, tedious, and expensive. The proposed system is designed as decision support system for medical doctors to classify the severity of knee OA. A method has been proposed here to localize a joint space area for OA and then classify it in 4 steps to classify OA into KL-Grade 0, KL-Grade 1, KL-Grade 2, KL-Grade 3 and KL-Grade 4, which are preprocessing, segmentation, feature extraction, and classification. In this proposed system, right and left knee detection was performed by employing the Contrast-Limited Adaptive Histogram Equalization (CLAHE) and the template matching. The Gabor kernel, row sum graph and moment methods were used to localize the junction space area of knee. CLAHE is used for preprocessing step, i.e.to normalize the varied intensities. The segmentation process was conducted using the Gabor kernel, template matching, row sum graph and gray level center of mass method. Here GLCM (contrast, correlation, energy, and homogeinity) features were employed as training data. Overall, 50 data were evaluated for training and 258 data for testing. Experimental results showed the best performance by using gabor kernel with parameters α=8, θ=0, Ψ=[0 π/2], γ=0,8, N=4 and with number of iterations being 5000, momentum value 0.5 and α0=0.6 for the classification process. The run gave classification accuracy rate of 93.8% for KL-Grade 0, 70% for KL-Grade 1, 4% for KL-Grade 2, 10% for KL-Grade 3 and 88.9% for KL-Grade 4.

Sound Modeling of Javanese Traditional Music Instrument

A Gamelan set consists of several groups of different instruments. One of the groups is called Balungan. Gamelan is contructed manually by hand with simple tools so it is very hard to find two gamelan sets are totally identical. In this research we propose to construct gamelan models. The main target of this research is creating Gamelan Frequency Modeling. We propose two Frequency Balungan Models, the first model is using average value, and the other is using average value in the most dense area.

Feature-points nearest neighbor clustering on 3D face models

Defining motion area on the face of 3D virtual character starts with the mapping of skeleton movement. Every animated character requires special handling based on the characteristics of the size and location of the bone to support producing facial expressions correctly. This process is often done specifically for each face model to be used. This research tried to use a marker-based motion capture data as a reference for the automation of generating clusters adaptively in the face of 3D characters. Each vertex which forming expression on the faces of the 3D models selected as centroids of cluster and representation a motion area whose numbers will correspond with the number of feature-point markers of motion capture data. Clustering process is done with the synthesis of modified nearest neighbor approach with the feature-point value. The results obtained were able to demonstrate a clustering process for generating motion area in a variety of 3D face model.

Adaptive threshold for background subtraction in moving object detection using Fuzzy C-Means clustering

Background subtraction is the important part of moving object detection. The problem of background subtraction is threshold selection strategy. This paper proposed a Fuzzy C-Means (FCM) algorithm to produce an adaptive threshold for background subtraction in moving object detection. To evaluate the performance, FCM were compared against standard Otsu algorithm as threshold selection strategy. Mean Square Error (MSE) and Peak Signal Noise Ratio (PSNR) was used to measure the performance. Based on the experiment, the MSE of FCM is lower than MSE of Otsu and PSNR of FCM is higher than PSNR of Otsu. The result proved that FCM is promising to classify the pixels as foreground or background in moving object detection.

Multi behavior NPC coordination using fuzzy coordinator and Gaussian distribution

Nowadays, Artificial Intelligence (AI) techniques have an important role in modern computer games especially to make NPCs in games more human-like, believable and natural. In a real time warfare game, coordination of NPC troops can create a deeper sense of immersion. Anyhow, multi behavior NPC troops have intelligence for selecting their behavior itself which sometimes does not appropriate to accomplish a team objective. Therefore, in this research we propose to develop smart agent for team coordination to produce better team behavior. Fuzzy coordinator method is employed for team coordination based on smart agent as a leader in attacking scenario. The smart agent coordinates a team strategy with uncoordinated action from the NPC to produce coordinated action for each NPC. Gaussian distribution is used to make variation action of each NPC more natural and unpredictable. The experiment demonstrates the smart agent with fuzzy coordinator who becomes the leader can analyze attack time remaining, health of each NPC and enemy to decide which one has to offense or defense in battle by change decision of each NPC troop to select action that has contribution for team.

Large Scale Text Classification Using Map Reduce and Naive Bayes Algorithm for Domain Specified Ontology Building

Internet that covers a large information gives an opportunity to obtain knowledge from it. Internet contains large unstructured and unorganized data such as text, video, and image. Problems arise on how to organize large amount of data and obtain a useful information from it. This information can be used as knowledge in the intelligent computer system. Ontology as one of knowledge representation covers a large area topic. For constructing domain specified ontology, we use large text dataset on Internet and organize it into specified domain before ontology building process is done. We try to implement naive bayes text classifier using map reduce programming model in our research for organizing our large text dataset. In this experiment, we use animal and plant domain article in Wikipedia online encyclopedia as our dataset. Our proposed method can achieve highest accuracy with score about 98.8%. This experiment shows that our proposed method provides a robust system and good accuracy for classifying document into specified domain in preprocessing phase for domain specified ontology building.

Multi-parameter dynamic difficulty game's scenario using Box-Muller of Gaussian distribution

Scenario is an important aspect in a game. It controls players experience according to the scenario that has been composed. Diversity design of scenario and unpredicted event make the game more challenging. This paper investigates the usage of multi-parameter Box-Muller method of Gaussian distribution in adjusting dynamically game scenario. Those parameters are mean (μ) and standard deviation (σ). Scenario is designed at cave stage of pedagogical game Reog Ponorogo using mathematics problems as games challenge. Challenges are defined as six categories cognitive domain of Bloom taxonomy. Those categories are: knowledge, comprehension, application, analysis, synthesis, and evaluation. Problem domain includes the following: sequences and series, probability and mathematical logic. Box-Muller method is used to select five of ten available problems at random, and Gaussian distribution was used to dynamically adjusting difficulty level of problems in order to match players skill.

Multi agent with multi behavior based on particle swarm optimization (PSO) for crowd movement in fire evacuation

A Simulation of human behavior are challenge topics of research in computer intelligent that has many benefits, such as to create a simulation of evacuation plan in a building when a disaster happens, like a fire and an earthquake. For evacuation of the building simulation of human behavior is needed to know which path will be passed through by the crowd. This animation simulation will use particle swarm algorithm optimization as the algorithm based with multiple behaviors and multiple targets. In this simulation, we add the behaviour of avoiding crashed between human and the algorithm modification is done which is a leader character is added. The leader behavior will lead the other agent to get out of the room. This agent based simulation movement will simulate movement in a room when an alarm signal is given, then the agent will get out of the room either individually or in groups. In this research we will use three scenario.We will compared the use of multiple target than single target and the use of leader follower behavior in any different number of agents.From the test result is obtained that the use of multiple target is much better result than use a single target and the behavior of the agent is depend on the movement of the crowd. Utilization of multibehavior with the leader characteristic who direct the other agent to reach target is more useful because it will reach the target more faster but the number of agents will affect the optimal number of leader needed.

WordNet Editor to Refine Indonesian Language Lexical Database

This paper describes an approach for editing Indonesian Language Lexical Database especially noun category and its relations. The purpose of this editor is to refine Indonesian Lexical Database that was developed in our previous researches. The visualization of the editor is using graph library with some modifications and additions. Furthermore, this editor will be web based so that everyone can participate to improve Indonesian Language Lexical Database. There is an administrator role that had to accept or reject any suggestion for the changes suggested by any member. We believe that this editing approach can also be used to improve WordNet developed in other languages.