Yoyon K. Suprapto

Facial emotional expressions recognition based on Active Shape Model and Radial Basis Function Network

Facial emotional expressions recognition (FEER) is important research fields to study how human beings reflect to environments in affective computing. With the rapid development of multimedia technology especially image processing, facial emotional expressions recognition researchers have achieved many useful result. If we want to recognize the humans emotion via the facial image, we need to extract features of the facial image. Active Shape Model (ASM) is one of the most popular methods for facial feature extraction. The accuracy of ASM depends on several factors, such as brightness, image sharpness, and noise. To get better result, the ASM is combined with Gaussian Pyramid. In this paper we propose a facial emotion expressions recognizing method based on ASM and Radial Basis Function Network (RBFN). Firstly, facial feature should be extracted to get emotional information from the region, but this paper use ASM method by the reconstructed facial shape. Second stage is to classify the facial emotion expressions from the emotional information. Finally get the model which is matched with the facial feature outline after several iterations and use them to recognize the facial emotional expressions by using RBFN. The experimental result from RBFN classifiers show a recognition accuracy of 90.73% for facial emotional expressions using the proposed method.

FFT-based features selection for Javanese music note and instrument identification using support vector machines

Most automatic music transcription research is related with Western music, and still less for the Javanese gamelan music. In this paper, we proposed a method for the features extraction, selection, and identification of gamelan note and the proper instrument. It was an approach based on Fast Fourier Transform (FFT), and support vector machines (SVMs) for note and instrument identification. We selected four spectral features (spectral centroid, two spectral rolloff, and fundamental frequency) as input for SVM. Experimental results show that fundamental frequency, spectral centroid, and spectral rolloff can be used to distinguish gamelan instrument with accuracy or recognition rate more than 95%.

Gamelan music onset detection using Elman Network

Gamelan, one of Indonesias traditional music instruments, generates signals that have variations in terms of fundamental frequency, amplitude, and signal envelope, due to its handmade construction and playing style. Therefore onset detection which is crucial for gamelan music analysis; undergoes several shortcomings using spectral and temporal features. This paper investigates the implementation of machine learning approach to understand statistical variations contained in gamelan signals which are relevant to onsets. The method uses Elman Network which consists of one hidden layer. Input units came from the power spectrogram and its positive first order difference of the signals as well as the context units from the output of each hidden unit one step back in time. The spectrogram was built using Short-time Fourier Transform and was converted into the log of Mel scale. A fixed threshold was used to select among the local peaks and the result is considered as binary classification of the signal at each time instant. The network was trained on a set of gamelan signals consists of synthetic and real recording data of single instrument playing. The performance gained 93% of F-measure.

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.

Music transcription of Javanese Gamelan using Short Time Fourier Transform (STFT)

Notation guide of traditional music such as Gamelan is very necessary for playing this instrument. There is a serious lack of artists who can add notation to represent gamelan, that cause arrangement research happens, the act of arranging and adapting a piece of gamelan notation. In this case, music transcription is the conversion process of music signal into numbers or symbols representation. The existence of this transcription research, the music can be played back using the results of transcription which used as a music notation guide in the play Gamelan. Process of transcription in this research use STFT (Short Time Fourier Transform) method. The results obtained in this paper were envelope signal from separation process the music signals into some channels of music notation in accordance with the fundamental frequency range of each music notation.

The optimization of the weblog central cluster using the genetic K-means algorithm

Clustering is part of data mining. Clustering is used to group objects so that one group has the same characteristics. K-means widely used because it is relatively easy to use. However K-means has shortcomings. K-means depends on the initial centroid. Selection of initial centroid done randomly so that the cluster formed is often not optimal. The clustering results are sometimes good and sometimes bad. In this research, the Genetic K-means Algorithm is used to improve K-means method. Genetic algorithm method is used to find the initial centroid. The initial centroid will be used by K-means. So K-means can get the optimal cluster. Cluster results is validated by SSW (Sum of Square within Cluster) and SI (Silhouette Index). SSW values by Genetic K-means Algorithm amounted 1,648,150,772.8 and K-means amounted 2.390.800.216,39. In this research, it was found that Genetic K-means Algorithm creates a homogenous cluster of 45% better than the K-means. So Genetic K-means Algorithm more accurate than K-means in determining patterns of data.

Ink bleed-through binarization of Javanese handwritten ancient document using local adaptive threshold based on local class width

Ink bleed-through is one of the degradations that usually appear in digitalization of ancient document. It makes the main text strongly hard to read. Consequently, binarization process is needed to split the main text and the background. Meanwhile, the popular binarization technique is thresholding and the common thresholding method does not work well to removes ink bleed-through degradation. Accordingly, we proposed new method using local adaptive threshold based on local class width. It has been done experimentally using Javanese handwritten ancient document named Babok Kalamadi, one of the collections in State Museum of Mpu Tantular, Sidoarjo, East Java, Indonesia. The document has ink bleed-through degradation and the proposed method achieves F-Measure and DRD that are: 91.33 and 3.58, 16.79 dB PSNR with 1361.17 MSE from the ground-truth measurement.

Texture Detection for Letter Carving Segmentation of Ancient Copper Inscriptions

As relics of history, ancient copper inscriptions are found in many countries. Information in the image or letter forms contained on copper ancient inscription has a very high value. The age and environmental factors caused damage to the surface of the inscription and also reduced the appearances of the image and letter. In this paper, we describe a novel segmentation methodology based on multi-texture features for ancient copper inscriptions which were severely damaged. The segmentation results of letters on ancient copper inscriptions by using the proposed method have an average accuracy of 90%. Based on these results, the proposed method is suitable for letter segmentation of the ancient copper inscriptions.

Segmentation and distribution of watershed using K-Modes clustering algorithm and Davies-Bouldin index based on geographic information system (GIS)

The watershed rehabilitation success rate have not been up, is the result of policies in watershed rehabilitation strategies that are less precise. From the above problems, we need a study that can provide a reference or any other alternative in determining priority watersheds to be rehabilitated, one through data mining. This paper uses a case study of Watershed data which are grouped using K-modes clustering algorithm based on its characteristics parameters. Watershed groupped using K-modes clustering then optimized using Davies-Bouildin Index (DBI) to get the number of clusters with the optimal level of similarity and visualized using GIS to obtain distribution maps. From trial on the Watershed of Tondano It was known that the cluster number four (4) is the optimal cluster number with an average DBI value of 0.672778, or 19.93%. The clustering results show that the wateshed in cluster 3 with 332 watershed which mostly scattered in the South Minahasa (24.7%) is a critical watershed compared to other clusters. the result of the clustering process is not much different or 90.64% similar when compared to the calculation of the watershed manually, that can be used as alternative to other reference in planning the rehabilitation of the watershed.

Determination of priority parameter for classification of poverty using Chi-square method and Crammer's V correlation

Poverty was a problem that faced by many developing countries, especially Indonesia. One way to resolve the issue of poverty through social assistance provided by the government. Besides that, knowing the factors affecting poverty in the region was also important to determine the strategic plan to reduce poverty in Indonesia. Data mining approach was used to determine the classification model. The classification method using Naïve Bayes algorithm while the process of determining the parameters using Chi-square and Crammers V correlation. The result of this experiment show that the parameter that has the most relation between the level of poverty was the number of household members (vART), and the smallest correlation was the fuel for cooking (vFUEL). Furthermore, the results of accuracy from Naïve Bayes have been obtained about 61.47% or better 0.54% when compared to the number of initial parameters. It means that the data give an explanation about 61.47% of the poverty level.