Romi Fadillah Rahmat

Skin color segmentation using multi-color space threshold

Skin color segmentation has proven to be useful in various application, such as face detection, hand gesture analysis, image content filtering, etc. There are some challenges on skin-color detection, such as wide variety of illumination condition and skin-like color objects appear in the background of image. Therefore, a method is required to cope with illumination condition while detecting skin color. In our proposed method, we combine the chrominance channels of three color spaces, namely HSV, YCbCr and Normalized RGB, to produce better rate result for skin color segmentation. The result shows that the proposed method is able to detect skin pixel and get up to 91.05% correct result while being tested in ECU and HGR Dataset. This result is significantly higher than the other single color spaces.

Distributed autonomous Neuro-Gen Learning Engine for content-based document file type identification

A Distributed Autonomous Neuro-Gen Learning Engine (DANGLE) is proposed in this paper for file type identification. DANGLE is a machine learning tool designed to solve limitations of existing implementation of neural networks, namely excessive training time, fixed architecture and catastrophic forgetting. DANGLE consists of a Gene Regulatory Engine (GRE) and a Distributed Adaptive Neural Network (DANN). File type identification is one of the phases in computer forensics, especially document file type identification. File type identification is a process of knowing the format of a file to determine the real file type of the file. In this paper, it is shown that DANGLEs performance is better than both EFuNN and ECF in identifying file type. The proposed DANGLE is also capable of identifying document files with an accuracy of 94.33%.

Implementation of Player Position Monitoring for Tanjung Pura Palace Virtual Environment.

Previously we had developed 3d model in virtual environment for palace in Tanjug Pura Region. For enchachment and enrichment of this application in order to introduce historical place in Tanjung Pura region, such as Darul Aman Palace, Darussalam Palace and the other historical buildings, we developed player positioning monitoring feature and implement it on its virtual environment. In this research, the player position mapping is performed inside a mini-map in Tanjung Pura Palace virtual reality application. The result of this research shows that the implemented method can be used to track the location of the user Tanjung Pura Palace virtual reality application.

Finger Recognition as Interaction Media in Augmented Reality for Historical Buildings in Matsum and Kesawan Regions of Medan City

Cultural heritage refers to the objects or symbolic attributes which resembles the identity of a society inherited from the previous generations, and will be preserved for the future generations. Despite of the importance of the cultural heritage, the interest of modern society for historic building has been decreased, as the cultural heritage is not considered as the main priority of development by the local government. This condition increases the necessity to restore the interest of modern society for historic building by implementing augmented reality. In this research, the finger recognition is utilized as the media of interaction between the 3-D objects and User. The system will identify the structure of human hand and calculate the number of fingers detected from the image obtained from web camera, by using convex hull and convexity defects. The research shows that the distance required to obtain the best performance is 30 to 50 cm with adequate light condition, while the distance required to perform marker detection in augmented reality is between 2.5 to 5 m, with the camera angle of 40°.

Historical Game of Majapahit Kingdom based on Tactical Role-playing Game.

Majapahit was a kingdom centered in East Java, which once stood around year 1293 to 1500 C. Majapahit kingdom was the last Hindu-Buddhist kingdom that controlled Nusantara and is regarded as one of the greatest kingdom in Indonesia. The lack of modern entertainment content about the history of Majapahit kingdom made historical subject become less attractive. Therefore, we need a modern entertainment as one option to learn about the fascinating history of the kingdom of Majapahit. In this study the authors designed a video game about history of Majapahit kingdom with the genre of tactical role-playing game. Tactical role-playing game is a sub genre of role playing game by using system of turn-based strategy in every battle. In tactical role-playing game, players will take turns with the opponent and can only take action in their turn and each character will have an attribute and level as in role-playing game video game. This study used the A* algorithm to determine the movement direction of the unit and cut-out techniques in the making of animation. This study demonstrated that video games can be used as a media to learn about history.

Russian Character Recognition using Self-Organizing Map

The World Tourism Organization (UNWTO) in 2014 released that there are 28 million visitors who visit Russia. Most of the visitors might have problem in typing Russian word when using digital dictionary. This is caused by the letters, called Cyrillic that used by the Russian and the countries around it, have different shape than Latin letters. The visitors might not familiar with Cyrillic. This research proposes an alternative way to input the Cyrillic words. Instead of typing the Cyrillic words directly, camera can be used to capture image of the words as input. The captured image is cropped, then several pre-processing steps are applied such as noise filtering, binary image processing, segmentation and thinning. Next, the feature extraction process is applied to the image. Cyrillic letters recognition in the image is done by utilizing Self-Organizing Map (SOM) algorithm. SOM successfully recognizes 89.09% Cyrillic letters from the computer-generated images. On the other hand, SOM successfully recognizes 88.89% Cyrillic letters from the image captured by the smartphones camera. For the word recognition, SOM successfully recognized 292 words and partially recognized 58 words from the image captured by the smartphones camera. Therefore, the accuracy of the word recognition using SOM is 83.42%

Scheduling Diet for Diabetes Mellitus Patients using Genetic Algorithm

Diabetes Melitus (DM) is one of metabolic diseases which affects on productivity and lowers the human resources quality. This disease can be controlled by maintaining and regulating balanced and healthy lifestyle especially for daily diet. However, nowadays, there is no system able to help DM patient to get any information of proper diet. Therefore, an approach is required to provide scheduling diet every day in a week with appropriate nutrition for DM patients to help them regulate their daily diet for healing this disease. In this research, we calculate the number of caloric needs using Harris-Benedict equation and propose genetic algorithm for scheduling diet for DM patient. The results show that the greater the number of individuals, the greater the more the possibility of changes in fitness score approaches the best fitness score. Moreover, the greater the created generation, the more the opportunites to obtain best individual with fitness score approaching 0 or equal to 0.

Automatic Text Summarization for Indonesian Language Using TextTeaser

Text summarization is one of the solution for information overload. Reducing text without losing the meaning not only can save time to read, but also maintain the readers understanding. One of many algorithms to summarize text is TextTeaser. Originally, this algorithm is intended to be used for text in English. However, due to TextTeaser algorithm does not consider the meaning of the text, we implement this algorithm for text in Indonesian language. This algorithm calculates four elements, such as title feature, sentence length, sentence position and keyword frequency. We utilize TextRank, an unsupervised and language independent text summarization algorithm, to evaluate the summarized text yielded by TextTeaser. The result shows that the TextTeaser algorithm needs more improvement to obtain better accuracy.

Chinese chess character recognition using Direction Feature Extraction and backpropagation

Backpropagation and Direction Feature Extraction (DFE) are proposed in this paper for Chinese chess character recognition. Backpropagation is a feed-forward neural network algorithm designed for learning by examples namely by calculating errors and updating weights in each epoch. DFE is a feature extraction method by iterating and calculating the directons surrounding each pixel in the image to obtain the features. In this research, Chinese chess characters are recognized to obtain the correct amount of each chess character in a package. Due to the complex contour, stroke and pattern of Chinese chess characters, Chinese chess characters are difficult to be recognized by new learners. Both Backpropagation and DFE performance are capable in recognizing Chinese chess characters with good accuracy of 98% for various sets and it is also robust from transition, brightness, image noise and rotation up to 60°.

Noise Reduction in Breath Sound Files Using Wavelet Transform Based Filter

The development of science and technology in the field of healthcare increasingly provides convenience in diagnosing respiratory system problem. Recording the breath sounds is one example of these developments. Breath sounds are recorded using a digital stethoscope, and then stored in a file with sound format. This breath sounds will be analyzed by health practitioners to diagnose the symptoms of disease or illness. However, the breath sounds is not free from interference signals. Therefore, noise filter or signal interference reduction system is required so that breath sounds component which contains information signal can be clarified. In this study, we designed a filter called a wavelet transform based filter. The filter that is designed in this study is using Daubechies wavelet with four wavelet transform coefficients. Based on the testing of the ten types of breath sounds data, the data is obtained in the largest SNRdB bronchial for 74.3685 decibels.